{"id":3936,"date":"2026-05-28T18:20:48","date_gmt":"2026-05-28T16:20:48","guid":{"rendered":"https:\/\/auranexus.ai\/the-platform-is-not-the-problem\/"},"modified":"2026-06-25T09:17:17","modified_gmt":"2026-06-25T07:17:17","slug":"the-platform-is-not-the-problem","status":"publish","type":"post","link":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/","title":{"rendered":"The platform is not the problem"},"content":{"rendered":"\t\t<div data-elementor-type=\"wp-post\" data-elementor-id=\"3936\" class=\"elementor elementor-3936 elementor-3918\" data-elementor-post-type=\"post\">\n\t\t\t\t<div class=\"elementor-element elementor-element-481c04e e-flex e-con-boxed e-con e-parent\" data-id=\"481c04e\" data-element_type=\"container\" data-e-type=\"container\">\n\t\t\t\t\t<div class=\"e-con-inner\">\n\t\t\t\t<div class=\"elementor-element elementor-element-12100bb elementor-widget elementor-widget-html\" data-id=\"12100bb\" data-element_type=\"widget\" data-e-type=\"widget\" data-widget_type=\"html.default\">\n\t\t\t\t\t<!--\n=============================================================\nauraNexus.ai Blog | auraHub &middot; DSGVO und Sprachmodelle | v4.2 SEO\/GEO Optimized | Mai 2026\n=============================================================\nWORDPRESS SETUP:\nSEO-Titel: Selbst gehostete KI-Plattform: DSGVO sicher | auraHub\nMeta-Description: Wenn auraHub bei Ihnen l&auml;uft, entscheidet die Wahl des Sprachmodells &uuml;ber den Datenschutz. Drei Sicherheitsstufen f&uuml;r drei Anwendungsf&auml;lle.\nSlug: dsgvo-konforme-ki-plattform-aurahub-drei-sicherheitsstufen\nFokus-Keyword: DSGVO-konforme KI-Plattform\nCHANGELOG v4.2:\n- Fix: Read-Next-Karten Hover-Schrift bleibt jetzt dunkel lesbar (Theme-Override mit !important)\n- Neu: Media-Slot Hover-Zoom (4 Prozent scale auf Bilder, 1.5 Prozent auf SVG, plus weicherer Schlagschatten)\n- Neu: Count-up-Animation fuer tx-inline-num und tx-note-stat (IntersectionObserver, 1.3 Sekunden, kubisches Easing)\n- Neu: prefers-reduced-motion respektiert, Animationen deaktivieren bei Nutzerwunsch\n- Fallback: data-count + textContent, Zahlen bleiben sichtbar auch ohne JavaScript\nCHANGELOG v4.1:\n- Slot 1 mit Bild befuellt: Moderner Arbeitsplatz mit auraHub-Dashboard und Blick auf den Koelner Dom\n- Neuer Slot 2 ergaenzt: Foto Server im Rechenzentrum als Plattform-Visualisierung in Sektion 02\n- Slot 2 (Architektur-SVG) auf Slot 3 umnummeriert, Slot 3 (Drei-Stufen-SVG) auf Slot 4\n- 21 externe Links zu autoritativen Quellen (DSGVO-Gesetz, EU-Kommission, Microsoft, OpenAI, Anthropic, vLLM, LM Studio, Jan.ai, OpenID Connect)\n- 9 inline interne Links plus 4 im Read-Next-Bereich\n- Alle Aussagen aus dem internen Datenschutz-Briefing zu auraHub (Stand Mai 2026), nichts erfunden\n- Zielgruppen-Anpassung: pragmatisch fuer GF und Marketing, Fachbegriffe kurz erklaert (SCC, AVV, ZDR, OIDC)\n- Komponenten: tx-modes, tx-comparetable, tx-sectors, tx-limits, vier tx-media-slot (2 Foto, 2 SVG)\n- Strukturlogik: Strategie-Tiefenartikel mit Plattform-Modell-Trennung als Kernidee\nMEDIA-SLOTS (4, alle befuellt):\n- Slot 1: Hero in Sektion 01 | 16:9 | Foto Arbeitsplatz mit auraHub-Dashboard und Koelner Dom\n- Slot 2: Plattform-Visualisierung in Sektion 02 | 16:9 | Foto Server im Rechenzentrum mit moderner Netzwerk-Infrastruktur\n- Slot 3: Architektur-Diagramm in Sektion 02 | 16:9 | SVG (Plattform-Modell-Trennung)\n- Slot 4: Drei-Stufen-Diagramm in Sektion 03 | 16:9 | SVG (Sicherheitsstufen am Sprachmodell)\n=============================================================\n-->\n<script type=\"application\/ld+json\">\n{\n  \"@context\": \"https:\/\/schema.org\",\n  \"@graph\": [\n    {\n      \"@type\": \"BlogPosting\",\n      \"headline\": \"Die Plattform ist nicht das Problem\",\n      \"description\": \"Datenschutz bei selbst gehosteten KI-Plattformen. Wenn auraHub in der eigenen Infrastruktur l\u00e4uft, entscheidet die Wahl des Sprachmodells \u00fcber den Datenschutz. Drei Sicherheitsstufen f\u00fcr drei Anwendungsf\u00e4lle, plus die Premium-Option Zero Data Retention f\u00fcr regulierte Daten.\",\n      \"author\": {\n        \"@type\": \"Person\",\n        \"name\": \"Oliver Range\",\n        \"jobTitle\": \"Gr\u00fcnder auraNexus.ai\",\n        \"worksFor\": {\"@type\": \"Organization\", \"name\": \"auraNexus.ai\", \"url\": \"https:\/\/auranexus.ai\"}\n      },\n      \"publisher\": {\"@type\": \"Organization\", \"name\": \"auraNexus.ai\", \"url\": \"https:\/\/auranexus.ai\"},\n      \"datePublished\": \"2026-05-28\",\n      \"inLanguage\": \"de-DE\",\n      \"keywords\": \"DSGVO-konforme KI-Plattform, auraHub, selbst gehostete KI, Sprachmodell DSGVO, Zero Data Retention, Azure OpenAI EU, lokales LLM, SCC, AVV\"\n    },\n    {\n      \"@type\": \"FAQPage\",\n      \"mainEntity\": [\n        {\"@type\": \"Question\", \"name\": \"Was bedeutet selbst gehostete KI-Plattform?\", \"acceptedAnswer\": {\"@type\": \"Answer\", \"text\": \"Eine selbst gehostete KI-Plattform l\u00e4uft in Ihrer eigenen Infrastruktur, wahlweise on-premises, in Ihrer Private Cloud oder air-gapped (ohne Internetverbindung). Alle plattformseitigen Daten wie Prompts, Dokumente, Nutzerrechte und Protokolle bleiben physisch in Ihrem Haus. auraHub ist als selbst gehostete KI-Plattform konzipiert.\"}},\n        {\"@type\": \"Question\", \"name\": \"Welche Sicherheitsstufen bietet auraHub am Sprachmodell?\", \"acceptedAnswer\": {\"@type\": \"Answer\", \"text\": \"Drei Stufen. Stufe 1 nutzt ein lokales Sprachmodell ohne jeden externen Datenabfluss. Stufe 2 nutzt Azure OpenAI in einer EU-Region mit AVV \u00fcber Microsoft. Stufe 3 nutzt die direkten APIs von OpenAI oder Anthropic, abgesichert \u00fcber Standardvertragsklauseln. Die Stufe wird pro Anwendungsfall gew\u00e4hlt.\"}},\n        {\"@type\": \"Question\", \"name\": \"Was ist Zero Data Retention?\", \"acceptedAnswer\": {\"@type\": \"Answer\", \"text\": \"Zero Data Retention (ZDR) ist eine Premium-Option f\u00fcr regulierte Daten. Ein- und Ausgaben werden nach der Antwort nicht mehr gespeichert, nicht in Missbrauchs-Logs abgelegt und keinem menschlichen Review zugef\u00fchrt. Die Verarbeitung erfolgt rein im Arbeitsspeicher. Anthropic, Microsoft Azure und OpenAI bieten ZDR im Rahmen von Enterprise-Vertr\u00e4gen an.\"}},\n        {\"@type\": \"Question\", \"name\": \"Wird auf unseren Daten trainiert?\", \"acceptedAnswer\": {\"@type\": \"Answer\", \"text\": \"Nein. Bei API- und Business-Tarifen ist zugesichert, dass kein Training auf den \u00fcbermittelten Daten stattfindet. Dies ist ein zentraler Unterschied zu Consumer-Apps wie ChatGPT Free, bei denen Daten standardm\u00e4\u00dfig f\u00fcr Modelltraining genutzt werden k\u00f6nnen.\"}},\n        {\"@type\": \"Question\", \"name\": \"Wie lange werden Eingaben au\u00dferhalb des Hauses gespeichert?\", \"acceptedAnswer\": {\"@type\": \"Answer\", \"text\": \"Beim lokalen LLM gar nicht. Azure OpenAI EU speichert standardm\u00e4\u00dfig bis 30 Tage zur Missbrauchserkennung in einer EU-Region. OpenAI API speichert standardm\u00e4\u00dfig bis 30 Tage, Anthropic 7 Tage (seit September 2025 von 30 reduziert). Mit Zero Data Retention entf\u00e4llt jede Speicherung nach der Antwort.\"}},\n        {\"@type\": \"Question\", \"name\": \"Welche DSGVO-Bausteine sind abgedeckt?\", \"acceptedAnswer\": {\"@type\": \"Answer\", \"text\": \"Sechs Bausteine: Auftragsverarbeitungsvertrag nach Artikel 28, Drittlandtransfer abgesichert \u00fcber Standardvertragsklauseln (entf\u00e4llt bei lokal oder EU), kein Training auf Ihren Daten, Datenminimierung und kurze Retention, Open Source und Auditierbarkeit des Plattform-Quellcodes, granulare Zugriffskontrolle mit Rollen, Gruppenrechten und SSO \u00fcber OpenID Connect.\"}},\n        {\"@type\": \"Question\", \"name\": \"Welche Stufe passt zu welchem Anwendungsfall?\", \"acceptedAnswer\": {\"@type\": \"Answer\", \"text\": \"Hochsensible Mandanten- oder Patientendaten laufen \u00fcber ein lokales Sprachmodell. Regul\u00e4re Gesch\u00e4ftsdaten mit EU-Pflicht \u00fcber Azure OpenAI EU. Unkritische Workloads \u00fcber die Direkt-API von OpenAI oder Anthropic. Wo Cloud-Modelle auf besonders sch\u00fctzenswerte Daten treffen, sichert Zero Data Retention die Verarbeitung ohne jede Speicherung ab.\"}}\n      ]\n    }\n  ]\n}\n<\/script>\n<style>\n@import url('https:\/\/fonts.googleapis.com\/css2?family=Nunito+Sans:ital,opsz,wght@0,6..12,300;0,6..12,400;0,6..12,500;0,6..12,600;0,6..12,700;1,6..12,400&family=JetBrains+Mono:ital,wght@0,400;0,500;0,600;1,400&display=swap');\n.tx-article {\n  --canvas: #F5F7F8;\n  --canvas-deep: #EAEEF0;\n  --canvas-soft: #FAFBFC;\n  --ink: #2E3338;\n  --ink-soft: #4a5058;\n  --ink-muted: #7a8088;\n  --ink-faint: #b0b5bc;\n  --teal: #4BB5B1;\n  --teal-deep: #3a9591;\n  --teal-soft: rgba(75,181,177,0.08);\n  --teal-line: rgba(75,181,177,0.25);\n  --accent: #F27F5E;\n  --accent-deep: #d96a4a;\n  --accent-soft: rgba(242,127,94,0.08);\n  --rule: rgba(46,51,56,0.10);\n  --rule-soft: rgba(46,51,56,0.05);\n}\n.tx-article * { box-sizing: border-box; margin: 0; padding: 0; }\n.tx-article {\n  background: var(--canvas);\n  color: var(--ink) !important;\n  font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;\n  font-weight: 400; line-height: 1.6; font-size: 16px;\n  overflow-x: hidden; position: relative;\n  margin: 0 -9999px; padding: 0 9999px;\n}\n.tx-article h1, .tx-article h2, .tx-article h3, .tx-article h4, .tx-article p, .tx-article span, .tx-article div, .tx-article blockquote, .tx-article cite, .tx-article summary, .tx-article a, .tx-article li, .tx-article ul, .tx-article ol { color: inherit; }\n.tx-mono { font-family: 'JetBrains Mono', 'SF Mono', 'Menlo', 'Consolas', monospace; }\n.tx-hero { padding: 120px 0 80px; position: relative; border-bottom: 1px solid var(--rule); }\n.tx-hero-grid { max-width: 1280px; margin: 0 auto; padding: 0 32px; display: grid; grid-template-columns: minmax(0, 8fr) minmax(0, 4fr); gap: 48px; align-items: end; }\n.tx-hero-eyebrow { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--teal); font-weight: 600; margin-bottom: 32px; display: inline-flex; align-items: center; gap: 12px; }\n.tx-hero-eyebrow::before { content: ''; width: 32px; height: 1.5px; background: var(--teal); display: inline-block; }\n.tx-hero-h1 { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: clamp(48px, 8vw, 120px); line-height: 0.95; font-weight: 700; letter-spacing: -0.04em; color: var(--ink) !important; margin-bottom: 24px; }\n.tx-hero-h1 em { font-style: normal; font-weight: 700; color: var(--teal) !important; display: inline-block; }\n.tx-hero-deck { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: clamp(17px, 1.6vw, 19px); line-height: 1.55; font-weight: 400; color: var(--ink-soft); max-width: 38ch; }\n.tx-hero-meta { margin-top: 64px; padding-top: 24px; border-top: 1px solid var(--rule); display: flex; gap: 32px; flex-wrap: wrap; font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.06em; color: var(--ink-muted); font-weight: 500; text-transform: uppercase; }\n.tx-hero-meta strong { color: var(--ink); font-weight: 600; }\n.tx-body { max-width: 1280px; margin: 0 auto; padding: 80px 32px; display: grid; grid-template-columns: minmax(220px, 1fr) minmax(0, 8fr) minmax(220px, 3fr); gap: 48px; position: relative; }\n.tx-sidebar { position: sticky; top: 80px; align-self: start; height: fit-content; }\n.tx-toc-label { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 10px; letter-spacing: 0.18em; text-transform: uppercase; color: var(--ink-muted); font-weight: 600; margin-bottom: 16px; }\n.tx-toc { list-style: none; border-left: 1px solid var(--rule); }\n.tx-toc li { padding: 0; }\n.tx-toc a { display: block; padding: 10px 0 10px 16px; font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 13px; font-weight: 500; color: var(--ink-muted); text-decoration: none; border-left: 2px solid transparent; margin-left: -1px; transition: color 0.2s, border-color 0.2s; line-height: 1.4; }\n.tx-toc a:hover { color: var(--ink); border-left-color: var(--teal-line); }\n.tx-toc-num { font-family: 'JetBrains Mono', monospace; font-size: 10px; color: var(--ink-faint); margin-right: 8px; font-weight: 500; }\n.tx-main { max-width: 100%; min-width: 0; }\n.tx-margin { position: relative; }\n.tx-note { margin-bottom: 48px; padding: 16px 0 16px 16px; border-left: 2px solid var(--teal); font-size: 13px; line-height: 1.55; color: var(--ink-soft); font-weight: 400; }\n.tx-note-label { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 10px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--teal); font-weight: 600; margin-bottom: 8px; display: block; }\n.tx-note-quote { font-size: 17px; line-height: 1.4; font-weight: 600; color: var(--ink); letter-spacing: -0.01em; }\n.tx-note-quote em { font-style: normal; color: var(--accent); font-weight: 600; }\n.tx-note-stat { font-size: 56px; line-height: 1; font-weight: 300; color: var(--teal); letter-spacing: -0.04em; margin: 8px 0 12px; display: block; }\n.tx-note-stat-label { font-size: 12px; line-height: 1.5; color: var(--ink-soft); font-weight: 400; }\n.tx-section { margin-bottom: 80px; scroll-margin-top: 80px; }\n.tx-section:last-child { margin-bottom: 0; }\n.tx-section-marker { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--accent); font-weight: 600; margin-bottom: 12px; display: flex; align-items: center; gap: 12px; }\n.tx-section-marker::before { content: ''; width: 8px; height: 8px; background: var(--accent); border-radius: 50%; display: inline-block; }\n.tx-section h2 { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: clamp(28px, 3.4vw, 40px); line-height: 1.15; font-weight: 700; letter-spacing: -0.025em; color: var(--ink) !important; margin-bottom: 32px; max-width: 22ch; }\n.tx-section h2 em { font-style: normal; color: var(--teal); }\n.tx-section h3 { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 19px; font-weight: 600; margin: 40px 0 12px; color: var(--ink) !important; letter-spacing: -0.01em; }\n.tx-section p { font-size: 17px; line-height: 1.75; color: var(--ink) !important; margin-bottom: 1.4em; font-weight: 400; max-width: 64ch; }\n.tx-section p strong { font-weight: 600; color: var(--ink) !important; }\n.tx-section a { color: var(--teal) !important; text-decoration: none; border-bottom: 1px solid var(--teal-line); transition: border-color 0.2s, color 0.2s; }\n.tx-section a:hover { border-bottom-color: var(--teal); color: var(--teal-deep) !important; }\n.tx-definition { margin: 32px 0 48px; padding: 28px 32px; background: var(--teal-soft); border-left: 3px solid var(--teal); border-radius: 4px; }\n.tx-definition-label { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--teal-deep); font-weight: 600; margin-bottom: 12px; }\n.tx-definition p { font-size: 16px; line-height: 1.6; color: var(--ink); margin: 0; max-width: 100%; }\n.tx-definition p strong { color: var(--teal-deep); }\n.tx-comparetable { margin: 48px 0; overflow-x: auto; }\n.tx-comparetable table { width: 100%; border-collapse: collapse; background: var(--canvas-soft); border: 1px solid var(--rule); border-radius: 6px; overflow: hidden; font-size: 14px; }\n.tx-comparetable th { background: var(--ink); color: var(--canvas); padding: 16px 20px; text-align: left; font-weight: 600; font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; letter-spacing: -0.005em; }\n.tx-comparetable th.tx-highlight { background: var(--teal); color: var(--canvas); }\n.tx-comparetable td { padding: 14px 20px; border-top: 1px solid var(--rule); color: var(--ink-soft); line-height: 1.5; vertical-align: top; }\n.tx-comparetable td:first-child { font-weight: 600; color: var(--ink); }\n.tx-comparetable td.tx-yes { color: var(--teal-deep); font-weight: 600; }\n.tx-comparetable td.tx-no { color: var(--accent-deep); font-weight: 600; }\n.tx-comparetable tr:hover { background: var(--canvas); }\n.tx-inline-numbers { display: grid; grid-template-columns: repeat(3, 1fr); gap: 32px; margin: 48px 0; padding: 32px 0; border-top: 1px solid var(--rule); border-bottom: 1px solid var(--rule); }\n.tx-inline-number { display: flex; flex-direction: column; }\n.tx-inline-num { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: clamp(48px, 6vw, 80px); line-height: 1; font-weight: 300; color: var(--teal); letter-spacing: -0.04em; margin-bottom: 12px; }\n.tx-inline-label { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.06em; color: var(--ink-soft); line-height: 1.5; font-weight: 500; }\n.tx-modes { display: grid; grid-template-columns: repeat(3, 1fr); gap: 16px; margin: 48px 0; }\n.tx-mode-card { padding: 24px 28px; background: var(--canvas-soft); border: 1px solid var(--rule); border-radius: 6px; position: relative; transition: all 0.2s; }\n.tx-mode-card:hover { border-color: var(--teal); background: var(--canvas); transform: translateY(-2px); }\n.tx-mode-key { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 10px; letter-spacing: 0.12em; color: var(--accent); font-weight: 600; margin-bottom: 8px; text-transform: uppercase; }\n.tx-mode-title { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 19px; font-weight: 600; margin-bottom: 8px; color: var(--ink); letter-spacing: -0.01em; }\n.tx-mode-output { font-size: 14px; color: var(--ink-soft); line-height: 1.6; font-weight: 400; }\n.tx-sectors { margin: 48px 0; }\n.tx-sector { display: grid; grid-template-columns: 200px 1fr; gap: 32px; padding: 24px 0; border-bottom: 1px solid var(--rule); align-items: start; }\n.tx-sector:last-child { border-bottom: none; }\n.tx-sector-tag { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.06em; text-transform: uppercase; color: var(--accent); font-weight: 600; line-height: 1.5; padding-top: 4px; }\n.tx-sector-content h3 { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 17px; font-weight: 600; margin-bottom: 6px; color: var(--ink); letter-spacing: -0.01em; }\n.tx-sector-content p { font-size: 14px; line-height: 1.65; color: var(--ink-soft); margin: 0; font-weight: 400; }\n.tx-limits { margin: 48px 0; list-style: none; }\n.tx-limits li { display: grid; grid-template-columns: 64px 1fr; gap: 24px; padding: 24px 0; border-top: 1px solid var(--rule); align-items: start; }\n.tx-limits li:last-child { border-bottom: 1px solid var(--rule); }\n.tx-limit-num { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 13px; font-weight: 600; color: var(--teal); padding-top: 4px; letter-spacing: 0.04em; }\n.tx-limit-text h3 { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 17px; font-weight: 600; margin-bottom: 6px; color: var(--ink); letter-spacing: -0.01em; }\n.tx-limit-text p { font-size: 14px; line-height: 1.65; color: var(--ink-soft); margin: 0; font-weight: 400; }\n.tx-cta { margin: 96px 0 64px; padding: 48px; background: var(--ink); border-radius: 8px; color: var(--canvas); position: relative; overflow: hidden; }\n.tx-cta::before { content: ''; position: absolute; top: 0; right: 0; width: 200px; height: 200px; background: radial-gradient(circle at top right, var(--teal) 0%, transparent 70%); opacity: 0.15; pointer-events: none; }\n.tx-cta-tag { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--teal); font-weight: 600; margin-bottom: 16px; }\n.tx-cta h2 { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: clamp(28px, 3.6vw, 40px); font-weight: 700; line-height: 1.15; color: var(--canvas); margin-bottom: 16px; max-width: 16ch; letter-spacing: -0.02em; }\n.tx-cta h2 em { font-style: normal; color: var(--teal); }\n.tx-cta p { font-size: 16px; line-height: 1.6; color: rgba(245,247,248,0.75); max-width: 56ch; margin: 0 0 32px 0; font-weight: 400; }\n.tx-cta-btn { display: inline-flex; align-items: center; gap: 12px; background: var(--teal) !important; color: var(--ink) !important; padding: 14px 28px; font-size: 14px; letter-spacing: 0.04em; font-weight: 700; text-decoration: none !important; border: none; border-radius: 4px; transition: all 0.2s; }\n.tx-cta-btn:hover { background: var(--canvas) !important; color: var(--ink) !important; text-decoration: none !important; transform: translateX(4px); }\n.tx-faq { margin: 80px 0; padding-top: 64px; border-top: 1px solid var(--rule); }\n.tx-faq-label { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--accent); font-weight: 600; margin-bottom: 12px; }\n.tx-faq h2 { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: clamp(26px, 3vw, 36px); font-weight: 700; line-height: 1.2; color: var(--ink); margin-bottom: 32px; letter-spacing: -0.02em; }\n.tx-faq h2 em { font-style: normal; color: var(--teal); }\n.tx-faq-item { border-bottom: 1px solid var(--rule); padding: 0; }\n.tx-faq-item:first-child { border-top: 1px solid var(--rule); }\n.tx-faq-q { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 16px; font-weight: 600; padding: 24px 48px 24px 32px; cursor: pointer; list-style: none; position: relative; color: var(--ink); line-height: 1.4; transition: color 0.2s, background 0.2s; letter-spacing: -0.005em; }\n.tx-faq-q::-webkit-details-marker { display: none; }\n.tx-faq-q::before { content: '?'; position: absolute; left: 0; top: 24px; font-family: 'JetBrains Mono', monospace; font-size: 18px; color: var(--teal); font-weight: 600; width: 24px; }\n.tx-faq-q::after { content: '+'; position: absolute; right: 16px; top: 50%; transform: translateY(-50%); font-family: 'JetBrains Mono', monospace; font-size: 20px; color: var(--ink-muted); font-weight: 400; transition: transform 0.3s, color 0.2s; }\n.tx-faq-item details[open] .tx-faq-q::after { transform: translateY(-50%) rotate(45deg); color: var(--teal); }\n.tx-faq-item:hover .tx-faq-q { color: var(--teal); background: var(--canvas-soft); }\n.tx-faq-a { font-size: 14px; line-height: 1.75; color: var(--ink-soft); font-weight: 400; padding: 0 48px 28px 32px; max-width: 64ch; }\n.tx-faq-a a { color: var(--teal); border-bottom: 1px solid var(--teal-line); text-decoration: none; }\n.tx-author { margin: 80px 0; padding: 32px; background: var(--canvas-soft); border: 1px solid var(--rule); border-radius: 8px; display: grid; grid-template-columns: 64px 1fr; gap: 24px; align-items: start; }\n.tx-author-avatar { width: 64px; height: 64px; background: var(--teal); color: var(--canvas); display: flex; align-items: center; justify-content: center; font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 22px; font-weight: 700; border-radius: 50%; letter-spacing: -0.02em; }\n.tx-author-handle { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.04em; color: var(--ink-muted); margin-bottom: 4px; font-weight: 500; }\n.tx-author-name { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 18px; font-weight: 700; margin-bottom: 4px; color: var(--ink); letter-spacing: -0.01em; }\n.tx-author-role { font-size: 13px; color: var(--ink-soft); margin-bottom: 12px; font-weight: 400; }\n.tx-author-bio { font-size: 13px; line-height: 1.65; color: var(--ink-soft); font-weight: 400; }\n.tx-author-bio a { color: var(--teal); border-bottom: 1px solid var(--teal-line); text-decoration: none; }\n.tx-readnext { margin: 64px 0; padding-top: 48px; border-top: 1px solid var(--rule); }\n.tx-readnext-label { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--ink-muted); font-weight: 600; margin-bottom: 24px; }\n.tx-readnext-grid { display: grid; grid-template-columns: repeat(2, 1fr); gap: 16px; }\n.tx-readnext-card { display: block; padding: 24px; background: var(--canvas-soft); border: 1px solid var(--rule); border-radius: 6px; text-decoration: none !important; color: var(--ink) !important; transition: all 0.2s; }\n.tx-readnext-card:hover { border-color: var(--teal); background: var(--canvas) !important; transform: translateY(-2px); color: var(--ink) !important; text-decoration: none !important; }\n.tx-readnext-card:hover .tx-readnext-title { color: var(--ink) !important; }\n.tx-readnext-card:hover .tx-readnext-cat { color: var(--accent) !important; }\n.tx-readnext-cat { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 10px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--accent) !important; font-weight: 600; margin-bottom: 8px; }\n.tx-readnext-title { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 15px; line-height: 1.4; font-weight: 600; letter-spacing: -0.005em; color: var(--ink) !important; }\n.tx-media-slot { margin: 48px 0; padding: 0; position: relative; }\n.tx-media-placeholder { position: relative; width: 100%; background: var(--canvas-soft); border: 2px dashed var(--rule); border-radius: 8px; display: flex; flex-direction: column; align-items: center; justify-content: center; gap: 10px; padding: 32px; text-align: center; transition: border-color 0.2s, background 0.2s; }\n.tx-media-placeholder:hover { border-color: var(--teal-line); background: var(--canvas); }\n.tx-media-slot[data-aspect=\"16-9\"] .tx-media-placeholder { aspect-ratio: 16 \/ 9; }\n.tx-media-slot[data-aspect=\"3-2\"] .tx-media-placeholder { aspect-ratio: 3 \/ 2; }\n.tx-media-slot[data-aspect=\"1-1\"] .tx-media-placeholder { aspect-ratio: 1 \/ 1; }\n.tx-media-label { font-family: 'JetBrains Mono', 'SF Mono', monospace; font-size: 11px; letter-spacing: 0.16em; text-transform: uppercase; color: var(--teal); font-weight: 600; }\n.tx-media-hint { font-size: 14px; color: var(--ink-muted); font-weight: 500; font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; max-width: 36ch; line-height: 1.45; }\n.tx-media-slot img, .tx-media-slot video, .tx-media-slot svg { width: 100%; height: auto; display: block; border-radius: 8px; transition: transform 0.7s cubic-bezier(0.22, 1, 0.36, 1), box-shadow 0.4s ease; will-change: transform; box-shadow: 0 4px 16px rgba(46,51,56,0.06); }\n.tx-media-slot:hover img, .tx-media-slot:hover video { transform: scale(1.04); box-shadow: 0 12px 32px rgba(46,51,56,0.18); }\n.tx-media-slot:hover svg { transform: scale(1.015); box-shadow: 0 10px 28px rgba(46,51,56,0.14); }\n.tx-media-caption { font-family: 'Avenir Next', 'Avenir', 'Nunito Sans', sans-serif; font-size: 13px; line-height: 1.5; color: var(--ink-muted); margin-top: 14px; font-style: italic; text-align: center; }\n.tx-inline-num { display: inline-block; }\n@keyframes tx-num-rise { from { opacity: 0; transform: translateY(16px); } to { opacity: 1; transform: translateY(0); } }\n.tx-inline-num.tx-num-revealed { animation: tx-num-rise 0.6s cubic-bezier(0.22, 1, 0.36, 1) both; }\n.tx-note-stat.tx-num-revealed { animation: tx-num-rise 0.6s cubic-bezier(0.22, 1, 0.36, 1) both; }\n@media (prefers-reduced-motion: reduce) {\n  .tx-media-slot img, .tx-media-slot video, .tx-media-slot svg { transition: none; }\n  .tx-media-slot:hover img, .tx-media-slot:hover video, .tx-media-slot:hover svg { transform: none; }\n  .tx-inline-num.tx-num-revealed, .tx-note-stat.tx-num-revealed { animation: none; }\n}\n@media (max-width: 1100px) {\n  .tx-body { grid-template-columns: minmax(0, 1fr); padding: 48px 24px; gap: 0; }\n  .tx-sidebar { display: none; }\n  .tx-margin { display: none; }\n  .tx-hero-grid { grid-template-columns: 1fr; gap: 32px; padding: 0 24px; }\n  .tx-hero { padding: 80px 0 48px; }\n  .tx-modes { grid-template-columns: 1fr; }\n  .tx-readnext-grid { grid-template-columns: 1fr; }\n  .tx-sector { grid-template-columns: 1fr; gap: 8px; }\n  .tx-cta { padding: 32px 24px; }\n  .tx-author { grid-template-columns: 1fr; gap: 16px; }\n  .tx-inline-numbers { grid-template-columns: 1fr; gap: 24px; }\n  .tx-faq-q { padding: 20px 36px 20px 28px; }\n  .tx-faq-a { padding: 0 36px 20px 28px; }\n  .tx-comparetable { font-size: 12px; }\n  .tx-comparetable th, .tx-comparetable td { padding: 10px 12px; }\n  .tx-media-placeholder { padding: 20px; gap: 8px; }\n}\n<\/style>\n<article class=\"tx-article\">\n  <section class=\"tx-hero\">\n    <div class=\"tx-hero-grid\">\n      <div>\n        <div class=\"tx-hero-eyebrow\">auraHub \u00b7 GDPR and language models<\/div>\n        <h1 class=\"tx-hero-h1\">The platform is not the <em>problem<\/em>.<\/h1>\n      <\/div>\n      <div>\n        <p class=\"tx-hero-deck\">If auraHub runs on your own infrastructure, your data stays with you. Who decides what happens to prompts and responses is determined by your choice of language model. Three levels for three use cases.  <\/p>\n      <\/div>\n    <\/div>\n    <div class=\"tx-hero-grid\">\n      <div class=\"tx-hero-meta\">\n        <span><strong>Oliver Range<\/strong><\/span>\n        <span>auraNexus.ai \u00b7 May 28, 2026<\/span>\n        <span>9 min read<\/span>\n      <\/div>\n      <div><\/div>\n    <\/div>\n  <\/section>\n  <div class=\"tx-body\">\n    <aside class=\"tx-sidebar\">\n      <div class=\"tx-toc-label\">Sections<\/div>\n      <ul class=\"tx-toc\">\n        <li><a href=\"#problem\"><span class=\"tx-toc-num\">01<\/span>The Problem<\/a><\/li>\n        <li><a href=\"#architektur\"><span class=\"tx-toc-num\">02<\/span>Architecture<\/a><\/li>\n        <li><a href=\"#stufen\"><span class=\"tx-toc-num\">03<\/span>Three levels<\/a><\/li>\n        <li><a href=\"#speicherung\"><span class=\"tx-toc-num\">04<\/span>Retention periods<\/a><\/li>\n        <li><a href=\"#bausteine\"><span class=\"tx-toc-num\">05<\/span>GDPR building blocks<\/a><\/li>\n        <li><a href=\"#empfehlung\"><span class=\"tx-toc-num\">06<\/span>Recommendation<\/a><\/li>\n        <li><a href=\"#workflow\"><span class=\"tx-toc-num\">07<\/span>Workflow<\/a><\/li>\n      <\/ul>\n    <\/aside>\n    <main class=\"tx-main\">\n      <!-- 01 PROBLEM -->\n      <section class=\"tx-section\" id=\"problem\">\n        <div class=\"tx-section-marker\">01 \/ Problem<\/div>\n        <h2>When people say GDPR, they mean the platform. The <em>risk<\/em> sits in the model. <\/h2>\n        <div class=\"tx-definition\">\n          <div class=\"tx-definition-label\">Definition<\/div>\n          <p>A <strong>GDPR-compliant AI platform<\/strong> refers to an AI application whose platform runs entirely within the company\u2019s infrastructure and whose language model is integrated per use case within a tiered security architecture. The platform governs data storage, permissions, and auditability. The language model determines what happens to prompts and responses once they leave the infrastructure.  <\/p>\n        <\/div>\n        <p>In reviews with IT security and data protection officers, one question always comes up first: Does it run on our servers? As soon as the answer is yes, the platform is considered solved and the discussion jumps to licensing. That is understandable, but this is where the real risk is overlooked.  <\/p>\n        <p>An AI platform consists of two layers. One contains prompts, uploaded documents, user permissions, protocols, knowledge sources, and configuration. You can keep this layer entirely in-house, both technically and contractually. The other layer is the language model that generates the actual answer. This is where it is decided whether content leaves the company, which region it goes to, how long it is stored there, and whether a provider may train on it.    <\/p>\n        <p>The productive question is therefore not whether it runs on our systems, but which model sees which data at which level. This is exactly the question that can be answered in a structured way with <a href=\"https:\/\/auranexus.ai\/en\/aurahub\/\">auraHub as a self-hosted AI platform<\/a>, because platform and model are cleanly separated architecturally. The legal basis is provided by the <a href=\"https:\/\/dsgvo-gesetz.de\/\" target=\"_blank\" rel=\"noopener\">General Data Protection Regulation<\/a>, whose obligations apply differently depending on the level.  <\/p>\n      <\/section>\n      <!-- SLOT 1 -->\n      <figure class=\"tx-media-slot\" data-slot=\"1\" data-aspect=\"16-9\">\n        <img decoding=\"async\" src=\"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Moderner-Arbeitsplatz-mit-auraHub-Dashboard-und-Blick-auf-den-Koelner-Dom.png\" alt=\"Modern workplace with the auraHub dashboard on a monitor, with Cologne Cathedral in the background. The platform runs in the company\u2019s own infrastructure at the Frechen site near Cologne.\" loading=\"lazy\">\n        <figcaption class=\"tx-media-caption\">Workplace with the auraHub dashboard and a view of Cologne Cathedral. The platform runs in-house\u2014you see it every day. <\/figcaption>\n      <\/figure>\n      <div class=\"tx-inline-numbers\">\n        <div class=\"tx-inline-number\">\n          <span class=\"tx-inline-num\" data-count=\"2\">2<\/span>\n          <span class=\"tx-inline-label\">Layers<br\/>Platform and language model<\/span>\n        <\/div>\n        <div class=\"tx-inline-number\">\n          <span class=\"tx-inline-num\" data-count=\"3\">3<\/span>\n          <span class=\"tx-inline-label\">Security levels<br\/>per use case<\/span>\n        <\/div>\n        <div class=\"tx-inline-number\">\n          <span class=\"tx-inline-num\" data-count=\"6\">6<\/span>\n          <span class=\"tx-inline-label\">GDPR building blocks<br\/>at a glance<\/span>\n        <\/div>\n      <\/div>\n      <!-- 02 ARCHITEKTUR -->\n      <section class=\"tx-section\" id=\"architektur\">\n        <div class=\"tx-section-marker\">02 \/ Architecture<\/div>\n        <h2>Two layers. <em>A clear separation<\/em>.<\/h2>\n        <p>auraHub runs entirely in your own environment\u2014either on-premises on your servers, in your private cloud, or air-gapped, i.e., without an internet connection. This means all platform-side data remains physically with you. At this layer, there is no transfer to a third country, no cloud lock-in, and no shift into external responsibility.  <\/p>\n        <!-- SLOT 2 -->\n        <figure class=\"tx-media-slot\" data-slot=\"2\" data-aspect=\"16-9\">\n          <img decoding=\"async\" src=\"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Server-im-Rechenzentrum-mit-moderner-Netzwerk-Infrastruktur.png\" alt=\"Servers in a data center with modern network infrastructure. Symbolic image for auraHub\u2019s self-hosted platform layer within the company\u2019s own infrastructure.\" loading=\"lazy\">\n          <figcaption class=\"tx-media-caption\">The physical platform layer: servers in your own infrastructure or private cloud infrastructure on which auraHub runs directly.<\/figcaption>\n        <\/figure>\n        <h3>What stays with you<\/h3>\n        <p>Specifically, the following remains in your infrastructure: prompts and inputs, uploaded documents, user and permission management, protocols and logs, connected knowledge sources, and the entire platform configuration. From a data protection perspective, this layer is the non-critical part. You have full control, no third-country transfer at platform level, and no cloud lock-in.  <\/p>\n        <h3>What the language model sees<\/h3>\n        <p>As soon as a prompt needs to be processed, auraHub calls the configured language model. This is exactly where content may potentially leave the infrastructure\u2014depending on the level: not at all (local model), to an EU region (<a href=\"https:\/\/learn.microsoft.com\/de-de\/azure\/ai-services\/openai\/concepts\/models\" target=\"_blank\" rel=\"noopener\">Azure OpenAI Service<\/a> with EU data residency), or to the US via <a href=\"https:\/\/commission.europa.eu\/law\/law-topic\/data-protection\/international-dimension-data-protection\/standard-contractual-clauses-scc_en\" target=\"_blank\" rel=\"noopener\">Standard Contractual Clauses<\/a> (OpenAI or Anthropic directly). This level can be set per use case, not as a blanket setting for the entire system. You may be familiar with a comparable architecture from our <a href=\"https:\/\/auranexus.ai\/en\/how-aurair-redefines-investor-relations\/\">auraIR setup for Investor Relations<\/a>, which follows the same logic.   <\/p>\n        <!-- SLOT 3: ARCHITEKTUR-SVG -->\n        <figure class=\"tx-media-slot\" data-slot=\"3\" data-aspect=\"16-9\">\n          <svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" viewbox=\"0 0 1600 900\" role=\"img\" aria-label=\"Architektur-Diagramm: auraHub-Plattform in der Kundeninfrastruktur links, drei Sicherheitsstufen am Sprachmodell rechts, verbunden \u00fcber einen API-Pfeil.\">\n            <title>auraHub-Architektur: Plattform und Sprachmodell getrennt<\/title>\n            <rect width=\"1600\" height=\"900\" fill=\"#F5F7F8\"\/>\n            <!-- Box 1 Plattform -->\n            <rect x=\"60\" y=\"120\" width=\"640\" height=\"660\" rx=\"12\" fill=\"#FAFBFC\" stroke=\"#4BB5B1\" stroke-width=\"2\"\/>\n            <text x=\"100\" y=\"180\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"22\" font-weight=\"700\" fill=\"#2E3338\" letter-spacing=\"-0.5\">Ihre Infrastruktur<\/text>\n            <text x=\"100\" y=\"212\" font-family=\"JetBrains Mono, monospace\" font-size=\"13\" font-weight=\"500\" fill=\"#4BB5B1\" letter-spacing=\"2\">EBENE PLATTFORM<\/text>\n            <text x=\"100\" y=\"252\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#4a5058\">On-premises, Private Cloud oder air-gapped<\/text>\n            <!-- Plattform-Inhalte -->\n            <g font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"16\" fill=\"#2E3338\">\n              <g>\n                <rect x=\"100\" y=\"320\" width=\"260\" height=\"60\" rx=\"6\" fill=\"#EAEEF0\"\/>\n                <text x=\"120\" y=\"358\" font-weight=\"600\">Prompts und Eingaben<\/text>\n              <\/g>\n              <g>\n                <rect x=\"380\" y=\"320\" width=\"260\" height=\"60\" rx=\"6\" fill=\"#EAEEF0\"\/>\n                <text x=\"400\" y=\"358\" font-weight=\"600\">Hochgeladene Dokumente<\/text>\n              <\/g>\n              <g>\n                <rect x=\"100\" y=\"400\" width=\"260\" height=\"60\" rx=\"6\" fill=\"#EAEEF0\"\/>\n                <text x=\"120\" y=\"438\" font-weight=\"600\">Nutzer und Rechte<\/text>\n              <\/g>\n              <g>\n                <rect x=\"380\" y=\"400\" width=\"260\" height=\"60\" rx=\"6\" fill=\"#EAEEF0\"\/>\n                <text x=\"400\" y=\"438\" font-weight=\"600\">Protokolle und Logs<\/text>\n              <\/g>\n              <g>\n                <rect x=\"100\" y=\"480\" width=\"260\" height=\"60\" rx=\"6\" fill=\"#EAEEF0\"\/>\n                <text x=\"120\" y=\"518\" font-weight=\"600\">Wissensquellen<\/text>\n              <\/g>\n              <g>\n                <rect x=\"380\" y=\"480\" width=\"260\" height=\"60\" rx=\"6\" fill=\"#EAEEF0\"\/>\n                <text x=\"400\" y=\"518\" font-weight=\"600\">Konfiguration<\/text>\n              <\/g>\n            <\/g>\n            <!-- Bewertung Box 1 -->\n            <rect x=\"100\" y=\"610\" width=\"540\" height=\"120\" rx=\"6\" fill=\"rgba(75,181,177,0.08)\" stroke=\"rgba(75,181,177,0.25)\" stroke-width=\"1\"\/>\n            <text x=\"120\" y=\"648\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#3a9591\" letter-spacing=\"2\">BEWERTUNG<\/text>\n            <text x=\"120\" y=\"680\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">Volle Kontrolle, kein Drittlandtransfer auf<\/text>\n            <text x=\"120\" y=\"704\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">Plattformebene, kein Cloud-Zwang.<\/text>\n            <!-- Pfeil -->\n            <g>\n              <line x1=\"700\" y1=\"450\" x2=\"900\" y2=\"450\" stroke=\"#F27F5E\" stroke-width=\"3\"\/>\n              <polygon points=\"900,450 880,440 880,460\" fill=\"#F27F5E\"\/>\n              <text x=\"800\" y=\"430\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#F27F5E\" letter-spacing=\"2\" text-anchor=\"middle\">API-AUFRUF<\/text>\n              <text x=\"800\" y=\"480\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"13\" fill=\"#4a5058\" text-anchor=\"middle\">nach gew\u00e4hlter Stufe<\/text>\n            <\/g>\n            <!-- Box 2 Sprachmodell -->\n            <rect x=\"900\" y=\"120\" width=\"640\" height=\"660\" rx=\"12\" fill=\"#2E3338\" stroke=\"#F27F5E\" stroke-width=\"2\"\/>\n            <text x=\"940\" y=\"180\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"22\" font-weight=\"700\" fill=\"#F5F7F8\" letter-spacing=\"-0.5\">Sprachmodell-Ebene<\/text>\n            <text x=\"940\" y=\"212\" font-family=\"JetBrains Mono, monospace\" font-size=\"13\" font-weight=\"500\" fill=\"#F27F5E\" letter-spacing=\"2\">EBENE MODELL<\/text>\n            <text x=\"940\" y=\"252\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"rgba(245,247,248,0.75)\">Drei Sicherheitsstufen, pro Anwendungsfall einstellbar<\/text>\n            <!-- 3 Stufen Mini-Karten -->\n            <g font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#F5F7F8\">\n              <g>\n                <rect x=\"940\" y=\"310\" width=\"560\" height=\"120\" rx=\"6\" fill=\"rgba(245,247,248,0.06)\"\/>\n                <text x=\"960\" y=\"346\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" fill=\"#4BB5B1\" letter-spacing=\"2\">STUFE 1<\/text>\n                <text x=\"960\" y=\"378\" font-size=\"18\" font-weight=\"700\">Lokales Sprachmodell<\/text>\n                <text x=\"960\" y=\"408\" font-size=\"14\" fill=\"rgba(245,247,248,0.75)\">H\u00f6chste Sicherheit, Verarbeitung im Haus<\/text>\n              <\/g>\n              <g>\n                <rect x=\"940\" y=\"450\" width=\"560\" height=\"120\" rx=\"6\" fill=\"rgba(245,247,248,0.06)\"\/>\n                <text x=\"960\" y=\"486\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" fill=\"#4BB5B1\" letter-spacing=\"2\">STUFE 2<\/text>\n                <text x=\"960\" y=\"518\" font-size=\"18\" font-weight=\"700\">Azure OpenAI EU<\/text>\n                <text x=\"960\" y=\"548\" font-size=\"14\" fill=\"rgba(245,247,248,0.75)\">EU-Datenresidenz, AVV \u00fcber Microsoft<\/text>\n              <\/g>\n              <g>\n                <rect x=\"940\" y=\"590\" width=\"560\" height=\"120\" rx=\"6\" fill=\"rgba(245,247,248,0.06)\"\/>\n                <text x=\"960\" y=\"626\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" fill=\"#4BB5B1\" letter-spacing=\"2\">STUFE 3<\/text>\n                <text x=\"960\" y=\"658\" font-size=\"18\" font-weight=\"700\">OpenAI oder Anthropic direkt<\/text>\n                <text x=\"960\" y=\"688\" font-size=\"14\" fill=\"rgba(245,247,248,0.75)\">SCC-abgesichert, kein Training auf API-Daten<\/text>\n              <\/g>\n            <\/g>\n          <\/svg>\n          <figcaption class=\"tx-media-caption\">Platform and language model are separated: what stays with you, and what is determined by the chosen level. <\/figcaption>\n        <\/figure>\n      <\/section>\n      <!-- 03 STUFEN -->\n      <section class=\"tx-section\" id=\"stufen\">\n        <div class=\"tx-section-marker\">03 \/ Three levels<\/div>\n        <h2>Local, EU, or safeguarded. <em>You choose per use case<\/em>.<\/h2>\n        <p>The three levels do not differ in auraHub\u2019s feature set, but in what happens outside your organization. The choice is made individually for each use case and can vary between use cases without rebuilding the platform. <\/p>\n        <div class=\"tx-modes\">\n          <div class=\"tx-mode-card\">\n            <div class=\"tx-mode-key\">Level 1 \u00b7 Highest security<\/div>\n            <div class=\"tx-mode-title\">Local language model<\/div>\n            <div class=\"tx-mode-output\"><a href=\"https:\/\/docs.vllm.ai\/\" target=\"_blank\" rel=\"noopener\">vLLM<\/a>, <a href=\"https:\/\/lmstudio.ai\/\" target=\"_blank\" rel=\"noopener\">LM Studio<\/a>, or <a href=\"https:\/\/jan.ai\/\" target=\"_blank\" rel=\"noopener\">Jan.ai<\/a>. Data never leaves the infrastructure\u2014no external transfer, no external storage. Suitable for highly sensitive client and patient data.  <\/div>\n          <\/div>\n          <div class=\"tx-mode-card\">\n            <div class=\"tx-mode-key\">Level 2 \u00b7 EU data residency<\/div>\n            <div class=\"tx-mode-title\">Azure OpenAI EU<\/div>\n            <div class=\"tx-mode-output\">EU region such as Sweden or France. Processing in EU data centers; data remains in the EU. DPA via <a href=\"https:\/\/www.microsoft.com\/de-de\/trust-center\/privacy\/gdpr-overview\" target=\"_blank\" rel=\"noopener\">Microsoft<\/a>; no training on your data. Suitable for standard business data with an EU requirement.   <\/div>\n          <\/div>\n          <div class=\"tx-mode-card\">\n            <div class=\"tx-mode-key\">Level 3 \u00b7 SCC-safeguarded<\/div>\n            <div class=\"tx-mode-title\">OpenAI or Anthropic directly<\/div>\n            <div class=\"tx-mode-output\">Providers\u2019 direct API. Processing in the US, safeguarded via Standard Contractual Clauses. No training on API data under <a href=\"https:\/\/openai.com\/enterprise-privacy\/\" target=\"_blank\" rel=\"noopener\">OpenAI Enterprise Privacy<\/a> and <a href=\"https:\/\/www.anthropic.com\/legal\/commercial-terms\" target=\"_blank\" rel=\"noopener\">Anthropic Commercial Terms<\/a>. Suitable for non-critical workloads.   <\/div>\n          <\/div>\n        <\/div>\n        <h3>Quick terminology clarification<\/h3>\n        <p><strong>DPA<\/strong> refers to the data processing agreement under <a href=\"https:\/\/dsgvo-gesetz.de\/art-28-dsgvo\/\" target=\"_blank\" rel=\"noopener\">Article 28 GDPR<\/a>, in which the cloud provider commits to processing your data only on your behalf and according to your instructions. <strong>SCC<\/strong> stands for <a href=\"https:\/\/commission.europa.eu\/law\/law-topic\/data-protection\/international-dimension-data-protection\/standard-contractual-clauses-scc_en\" target=\"_blank\" rel=\"noopener\">Standard Contractual Clauses<\/a>, a set of contractual clauses issued by the European Commission to safeguard the transfer of personal data to third countries. With a local model, neither applies because no data leaves the organization. <\/p>\n        <!-- SLOT 4: DREI STUFEN SVG -->\n        <figure class=\"tx-media-slot\" data-slot=\"4\" data-aspect=\"16-9\">\n          <svg xmlns=\"http:\/\/www.w3.org\/2000\/svg\" viewbox=\"0 0 1800 1000\" role=\"img\" aria-label=\"Drei Sicherheitsstufen f\u00fcr das Sprachmodell: Lokales LLM, Azure OpenAI EU und OpenAI oder Anthropic direkt, jeweils mit Verarbeitungsort, Speicherung und Eignung.\">\n            <title>Drei Sicherheitsstufen f\u00fcr das Sprachmodell<\/title>\n            <rect width=\"1800\" height=\"1000\" fill=\"#F5F7F8\"\/>\n            <!-- Header -->\n            <text x=\"80\" y=\"100\" font-family=\"JetBrains Mono, monospace\" font-size=\"13\" font-weight=\"600\" fill=\"#F27F5E\" letter-spacing=\"3\">EBENE SPRACHMODELL<\/text>\n            <text x=\"80\" y=\"150\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"36\" font-weight=\"700\" fill=\"#2E3338\" letter-spacing=\"-1\">Drei Sicherheitsstufen zur Wahl<\/text>\n            <!-- Stufe 1 Lokal -->\n            <g>\n              <rect x=\"80\" y=\"220\" width=\"540\" height=\"700\" rx=\"12\" fill=\"#2E3338\"\/>\n              <text x=\"120\" y=\"280\" font-family=\"JetBrains Mono, monospace\" font-size=\"12\" font-weight=\"600\" fill=\"#4BB5B1\" letter-spacing=\"2.5\">STUFE 1<\/text>\n              <text x=\"120\" y=\"312\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"22\" font-weight=\"700\" fill=\"#F5F7F8\">H\u00f6chste Sicherheit<\/text>\n              <text x=\"120\" y=\"354\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"28\" font-weight=\"700\" fill=\"#4BB5B1\" letter-spacing=\"-0.5\">Lokales LLM<\/text>\n              <text x=\"120\" y=\"392\" font-family=\"JetBrains Mono, monospace\" font-size=\"13\" fill=\"rgba(245,247,248,0.65)\">vLLM \u00b7 LM Studio \u00b7 Jan.ai<\/text>\n              <line x1=\"120\" y1=\"430\" x2=\"580\" y2=\"430\" stroke=\"rgba(245,247,248,0.2)\" stroke-width=\"1\"\/>\n              <text x=\"120\" y=\"470\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#4BB5B1\" letter-spacing=\"2\">DATEN<\/text>\n              <text x=\"120\" y=\"500\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#F5F7F8\">Verlassen die Infrastruktur nie,<\/text>\n              <text x=\"120\" y=\"524\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#F5F7F8\">vollst\u00e4ndig im Haus.<\/text>\n              <text x=\"120\" y=\"580\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#4BB5B1\" letter-spacing=\"2\">TRANSFER<\/text>\n              <text x=\"120\" y=\"610\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#F5F7F8\">Kein externer Transfer,<\/text>\n              <text x=\"120\" y=\"634\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#F5F7F8\">keine API nach au\u00dfen.<\/text>\n              <text x=\"120\" y=\"690\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#4BB5B1\" letter-spacing=\"2\">SPEICHERUNG<\/text>\n              <text x=\"120\" y=\"720\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#F5F7F8\">Keine externe Speicherung.<\/text>\n              <rect x=\"120\" y=\"800\" width=\"460\" height=\"80\" rx=\"6\" fill=\"rgba(75,181,177,0.15)\"\/>\n              <text x=\"140\" y=\"836\" font-family=\"JetBrains Mono, monospace\" font-size=\"10\" font-weight=\"600\" fill=\"#4BB5B1\" letter-spacing=\"2\">EIGNUNG<\/text>\n              <text x=\"140\" y=\"862\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"14\" fill=\"#F5F7F8\">Hochsensible Mandanten- und Patientendaten<\/text>\n            <\/g>\n            <!-- Stufe 2 Azure -->\n            <g>\n              <rect x=\"640\" y=\"220\" width=\"540\" height=\"700\" rx=\"12\" fill=\"#FAFBFC\" stroke=\"#4BB5B1\" stroke-width=\"2\"\/>\n              <text x=\"680\" y=\"280\" font-family=\"JetBrains Mono, monospace\" font-size=\"12\" font-weight=\"600\" fill=\"#F27F5E\" letter-spacing=\"2.5\">STUFE 2<\/text>\n              <text x=\"680\" y=\"312\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"22\" font-weight=\"700\" fill=\"#2E3338\">EU-Datenresidenz<\/text>\n              <text x=\"680\" y=\"354\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"28\" font-weight=\"700\" fill=\"#3a9591\" letter-spacing=\"-0.5\">Azure OpenAI EU<\/text>\n              <text x=\"680\" y=\"392\" font-family=\"JetBrains Mono, monospace\" font-size=\"13\" fill=\"#7a8088\">EU-Region \u00b7 Sweden, France<\/text>\n              <line x1=\"680\" y1=\"430\" x2=\"1140\" y2=\"430\" stroke=\"rgba(46,51,56,0.10)\" stroke-width=\"1\"\/>\n              <text x=\"680\" y=\"470\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#3a9591\" letter-spacing=\"2\">DATEN<\/text>\n              <text x=\"680\" y=\"500\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">Verarbeitung in EU-Rechen-<\/text>\n              <text x=\"680\" y=\"524\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">zentren, Daten bleiben in EU.<\/text>\n              <text x=\"680\" y=\"580\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#3a9591\" letter-spacing=\"2\">VERTRAG<\/text>\n              <text x=\"680\" y=\"610\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">AVV \u00fcber Microsoft,<\/text>\n              <text x=\"680\" y=\"634\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">kein Training auf Ihren Daten.<\/text>\n              <text x=\"680\" y=\"690\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#3a9591\" letter-spacing=\"2\">SPEICHERUNG<\/text>\n              <text x=\"680\" y=\"720\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">Bis 30 Tage, nur Missbrauch.<\/text>\n              <text x=\"680\" y=\"744\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">Mit ZDR: keine.<\/text>\n              <rect x=\"680\" y=\"800\" width=\"460\" height=\"80\" rx=\"6\" fill=\"rgba(75,181,177,0.10)\"\/>\n              <text x=\"700\" y=\"836\" font-family=\"JetBrains Mono, monospace\" font-size=\"10\" font-weight=\"600\" fill=\"#3a9591\" letter-spacing=\"2\">EIGNUNG<\/text>\n              <text x=\"700\" y=\"862\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"14\" fill=\"#2E3338\">Regul\u00e4re Gesch\u00e4ftsdaten mit EU-Pflicht<\/text>\n            <\/g>\n            <!-- Stufe 3 SCC -->\n            <g>\n              <rect x=\"1200\" y=\"220\" width=\"540\" height=\"700\" rx=\"12\" fill=\"#FAFBFC\" stroke=\"rgba(46,51,56,0.10)\" stroke-width=\"2\"\/>\n              <text x=\"1240\" y=\"280\" font-family=\"JetBrains Mono, monospace\" font-size=\"12\" font-weight=\"600\" fill=\"#F27F5E\" letter-spacing=\"2.5\">STUFE 3<\/text>\n              <text x=\"1240\" y=\"312\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"22\" font-weight=\"700\" fill=\"#2E3338\">SCC-abgesichert<\/text>\n              <text x=\"1240\" y=\"354\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"26\" font-weight=\"700\" fill=\"#2E3338\" letter-spacing=\"-0.5\">OpenAI \/ Anthropic<\/text>\n              <text x=\"1240\" y=\"392\" font-family=\"JetBrains Mono, monospace\" font-size=\"13\" fill=\"#7a8088\">Direkt-API der Anbieter<\/text>\n              <line x1=\"1240\" y1=\"430\" x2=\"1700\" y2=\"430\" stroke=\"rgba(46,51,56,0.10)\" stroke-width=\"1\"\/>\n              <text x=\"1240\" y=\"470\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#4a5058\" letter-spacing=\"2\">DATEN<\/text>\n              <text x=\"1240\" y=\"500\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">Verarbeitung in den USA,<\/text>\n              <text x=\"1240\" y=\"524\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">abgesichert \u00fcber SCC.<\/text>\n              <text x=\"1240\" y=\"580\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#4a5058\" letter-spacing=\"2\">TRAINING<\/text>\n              <text x=\"1240\" y=\"610\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">Kein Training auf API-Daten.<\/text>\n              <text x=\"1240\" y=\"690\" font-family=\"JetBrains Mono, monospace\" font-size=\"11\" font-weight=\"600\" fill=\"#4a5058\" letter-spacing=\"2\">SPEICHERUNG<\/text>\n              <text x=\"1240\" y=\"720\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">OpenAI 30 Tage,<\/text>\n              <text x=\"1240\" y=\"744\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"15\" fill=\"#2E3338\">Anthropic 7 Tage. Mit ZDR: keine.<\/text>\n              <rect x=\"1240\" y=\"800\" width=\"460\" height=\"80\" rx=\"6\" fill=\"rgba(242,127,94,0.10)\"\/>\n              <text x=\"1260\" y=\"836\" font-family=\"JetBrains Mono, monospace\" font-size=\"10\" font-weight=\"600\" fill=\"#d96a4a\" letter-spacing=\"2\">EIGNUNG<\/text>\n              <text x=\"1260\" y=\"862\" font-family=\"Avenir Next, Nunito Sans, sans-serif\" font-size=\"14\" fill=\"#2E3338\">Unkritische Workloads<\/text>\n            <\/g>\n          <\/svg>\n          <figcaption class=\"tx-media-caption\">Three security levels at the language model: local for highly sensitive data, EU for standard business data, SCC-safeguarded for non-critical workloads. <\/figcaption>\n        <\/figure>\n      <\/section>\n      <!-- 04 SPEICHERUNG -->\n      <section class=\"tx-section\" id=\"speicherung\">\n        <div class=\"tx-section-marker\">04 \/ Retention periods<\/div>\n        <h2>How long your inputs are stored <em>outside your organization<\/em>.<\/h2>\n        <p>Even if the language model is not trained on your data, inputs may be stored by the provider for a limited time, usually for abuse detection. Retention periods vary by provider and have been systematically decreasing in recent years. For particularly sensitive data, there is the premium option Zero Data Retention (ZDR), under which no storage takes place at all.  <\/p>\n        <div class=\"tx-comparetable\">\n          <table>\n            <thead>\n              <tr>\n                <th scope=\"col\">Option<\/th>\n                <th scope=\"col\">Content retention<\/th>\n                <th scope=\"col\">Training on content<\/th>\n              <\/tr>\n            <\/thead>\n            <tbody>\n              <tr>\n                <td>Local LLM<\/td>\n                <td class=\"tx-yes\">No external storage<\/td>\n                <td class=\"tx-yes\">No<\/td>\n              <\/tr>\n              <tr>\n                <td>Azure OpenAI EU (standard)<\/td>\n                <td>Up to 30 days, abuse detection only, in EU region<\/td>\n                <td class=\"tx-yes\">No<\/td>\n              <\/tr>\n              <tr>\n                <td>OpenAI API (standard)<\/td>\n                <td>Up to 30 days, then deleted<\/td>\n                <td class=\"tx-yes\">No<\/td>\n              <\/tr>\n              <tr>\n                <td>Anthropic API (standard)<\/td>\n                <td>7 days (reduced from 30 since September 2025)<\/td>\n                <td class=\"tx-yes\">No<\/td>\n              <\/tr>\n              <tr>\n                <td>With Zero Data Retention<\/td>\n                <td class=\"tx-yes\">No storage after the response, in-memory only<\/td>\n                <td class=\"tx-yes\">No<\/td>\n              <\/tr>\n            <\/tbody>\n          <\/table>\n        <\/div>\n        <h3>Premium option: Zero Data Retention<\/h3>\n        <p>For particularly sensitive processing\u2014such as finance and investor-relations data (as in our <a href=\"https:\/\/auranexus.ai\/en\/how-aurair-redefines-investor-relations\/\">auraIR setup for listed companies<\/a>)\u2014Anthropic offers Zero Data Retention as part of an enterprise agreement. Inputs and outputs are then no longer stored after the response, are not written to abuse logs, and are not subjected to human review. Processing takes place purely in memory.  <\/p>\n        <p>This level must be requested separately and is tied to a qualifying contract. <a href=\"https:\/\/learn.microsoft.com\/de-de\/legal\/cognitive-services\/openai\/data-privacy\" target=\"_blank\" rel=\"noopener\">Microsoft (Azure)<\/a> and <a href=\"https:\/\/openai.com\/enterprise-privacy\/\" target=\"_blank\" rel=\"noopener\">OpenAI<\/a> offer ZDR in a similar way for enterprise or Microsoft customer agreements. A practical note: with Anthropic, the results of security classifiers are still retained even under ZDR\u2014this should be considered in the risk model. <\/p>\n      <\/section>\n      <!-- 05 BAUSTEINE -->\n      <section class=\"tx-section\" id=\"bausteine\">\n        <div class=\"tx-section-marker\">05 \/ GDPR building blocks<\/div>\n        <h2>Six building blocks. <em>A consistent review<\/em>.<\/h2>\n        <p>The following six building blocks cover the GDPR-relevant obligations when using auraHub and can be answered consistently in any audit question.<\/p>\n        <div class=\"tx-sectors\">\n          <div class=\"tx-sector\">\n            <div class=\"tx-sector-tag\">Article 28<\/div>\n            <div class=\"tx-sector-content\">\n              <h3>Data processing agreement<\/h3>\n              <p>Available with every cloud provider and must be concluded before use. Under <a href=\"https:\/\/dsgvo-gesetz.de\/art-28-dsgvo\/\" target=\"_blank\" rel=\"noopener\">Art. 28 GDPR<\/a>, it governs that the provider processes your data only on your behalf. <\/p>\n            <\/div>\n          <\/div>\n          <div class=\"tx-sector\">\n            <div class=\"tx-sector-tag\">Third country<\/div>\n            <div class=\"tx-sector-content\">\n              <h3>Transfer safeguarded<\/h3>\n              <p>For US providers via <a href=\"https:\/\/commission.europa.eu\/law\/law-topic\/data-protection\/international-dimension-data-protection\/standard-contractual-clauses-scc_en\" target=\"_blank\" rel=\"noopener\">Standard Contractual Clauses<\/a>. Not applicable at all with a local or EU model, because no third-country transfer takes place. <\/p>\n            <\/div>\n          <\/div>\n          <div class=\"tx-sector\">\n            <div class=\"tx-sector-tag\">Training<\/div>\n            <div class=\"tx-sector-content\">\n              <h3>No training on your data<\/h3>\n              <p>Contractually assured for API and business plans. An important distinction from consumer apps, where inputs may be used for model training by default. <\/p>\n            <\/div>\n          <\/div>\n          <div class=\"tx-sector\">\n            <div class=\"tx-sector-tag\">Retention<\/div>\n            <div class=\"tx-sector-content\">\n              <h3>Data minimization<\/h3>\n              <p>Short retention periods on the provider side, with optional Zero Data Retention for regulated data. On the platform side, you control retention through your own configuration. <\/p>\n            <\/div>\n          <\/div>\n          <div class=\"tx-sector\">\n            <div class=\"tx-sector-tag\">Open Source<\/div>\n            <div class=\"tx-sector-content\">\n              <h3>Auditability<\/h3>\n              <p>The platform\u2019s source code is accessible; data flows are auditable. Important for internal audit, external audits, and for trust with customers, investors, and regulators. <\/p>\n            <\/div>\n          <\/div>\n          <div class=\"tx-sector\">\n            <div class=\"tx-sector-tag\">Access<\/div>\n            <div class=\"tx-sector-content\">\n              <h3>Granular permissions<\/h3>\n              <p>Role and group permissions, single sign-on via <a href=\"https:\/\/openid.net\/connect\/\" target=\"_blank\" rel=\"noopener\">OpenID Connect<\/a> (an open standard for centralized login), encrypted secrets. Who may see what is configurable down to the source level. <\/p>\n            <\/div>\n          <\/div>\n        <\/div>\n      <\/section>\n      <!-- 06 EMPFEHLUNG -->\n      <section class=\"tx-section\" id=\"empfehlung\">\n        <div class=\"tx-section-marker\">06 \/ Recommendation<\/div>\n        <h2>One platform. <em>Tiered model selection<\/em> per use case.<\/h2>\n        <p>The decisive lever for assessment in your organization is tiered model selection per use case. Highly sensitive cases run on a local model with no data leaving the organization. Standard business data runs via Azure in the EU. Non-critical workloads run via the direct API.   <\/p>\n        <p>Where cloud models touch particularly sensitive data, Zero Data Retention safeguards processing without any storage. This allows each use case to be set to the appropriate protection level individually, without having to move the entire operation to the maximum security level. This reduces costs where they do not buy additional protection, and concentrates effort where it has an effect.  <\/p>\n        <h3>What that means for your assessment<\/h3>\n        <p>Three questions are enough to arrive at the right level per use case. First: which data category is processed\u2014for example personal data under <a href=\"https:\/\/dsgvo-gesetz.de\/art-9-dsgvo\/\" target=\"_blank\" rel=\"noopener\">Article 9 GDPR<\/a>, trade secrets, or non-critical data? Second: which regulatory requirement applies\u2014for example professional confidentiality, an industry-specific requirement, or GDPR in general? Third: how frequently does the use case run, and does the operating model justify a local model, or is the cloud option sufficient? If you need an initial assessment of your cases, we support this as part of our <a href=\"https:\/\/auranexus.ai\/en\/contact\/\">AI consulting<\/a>.    <\/p>\n      <\/section>\n      <!-- 07 WORKFLOW -->\n      <section class=\"tx-section\" id=\"workflow\">\n        <div class=\"tx-section-marker\">07 \/ Workflow<\/div>\n        <h2>How to arrive at the right <em>level selection<\/em>.<\/h2>\n        <ul class=\"tx-limits\">\n          <li>\n            <span class=\"tx-limit-num\">01<\/span>\n            <div class=\"tx-limit-text\">\n              <h3>Collect use cases<\/h3>\n              <p>List between three and ten concrete use cases in which auraHub is to be used. Examples: summarizing press coverage, reviewing contract drafts, categorizing customer inquiries, pre-structuring patient letters. <\/p>\n            <\/div>\n          <\/li>\n          <li>\n            <span class=\"tx-limit-num\">02<\/span>\n            <div class=\"tx-limit-text\">\n              <h3>Assign a data category<\/h3>\n              <p>For each use case: does it involve particularly sensitive data under <a href=\"https:\/\/dsgvo-gesetz.de\/art-9-dsgvo\/\" target=\"_blank\" rel=\"noopener\">Article 9 GDPR<\/a>, standard personal data, trade secrets, or non-critical public data? This assignment directly determines the level selection. <\/p>\n            <\/div>\n          <\/li>\n          <li>\n            <span class=\"tx-limit-num\">03<\/span>\n            <div class=\"tx-limit-text\">\n              <h3>Select the level<\/h3>\n              <p>Level 1 local for highly sensitive data and professional confidentiality. Level 2 Azure EU for standard business data with an EU requirement. Level 3 direct API for non-critical workloads. Enable ZDR as soon as sensitive data touches cloud models.   <\/p>\n            <\/div>\n          <\/li>\n          <li>\n            <span class=\"tx-limit-num\">04<\/span>\n            <div class=\"tx-limit-text\">\n              <h3>Conclude contracts<\/h3>\n              <p>Conclude a DPA with all relevant providers (<a href=\"https:\/\/www.microsoft.com\/de-de\/trust-center\/privacy\/gdpr-overview\" target=\"_blank\" rel=\"noopener\">Microsoft<\/a>, <a href=\"https:\/\/openai.com\/enterprise-privacy\/\" target=\"_blank\" rel=\"noopener\">OpenAI<\/a>, <a href=\"https:\/\/www.anthropic.com\/legal\/commercial-terms\" target=\"_blank\" rel=\"noopener\">Anthropic<\/a>). Request ZDR explicitly where applicable. Review Standard Contractual Clauses where third-country transfer takes place.  <\/p>\n            <\/div>\n          <\/li>\n          <li>\n            <span class=\"tx-limit-num\">05<\/span>\n            <div class=\"tx-limit-text\">\n              <h3>Configuration and audit<\/h3>\n              <p>Configure auraHub per use case to the selected level. Assign roles and permissions according to the use case. Document data flows once in full, then review them regularly.  <\/p>\n            <\/div>\n          <\/li>\n        <\/ul>\n      <\/section>\n      <!-- CTA -->\n      <section class=\"tx-cta\">\n        <div class=\"tx-cta-tag\">$ Request consulting<\/div>\n        <h2>auraHub <em>in-house<\/em>.<\/h2>\n        <p>We support you from the initial assessment of your use cases through to productive go-live\u2014together with your data protection and IT stakeholders, in your infrastructure, with the right level selection per use case. <\/p>\n        <a href=\"https:\/\/auranexus.ai\/en\/contact\/\" class=\"tx-cta-btn\">Schedule an appointment \u2192<\/a>\n      <\/section>\n      <!-- FAQ -->\n      <section class=\"tx-faq\">\n        <div class=\"tx-faq-label\">FAQ \/ Reference<\/div>\n        <h2>Frequently asked questions<\/h2>\n        <div class=\"tx-faq-item\">\n          <details>\n            <summary class=\"tx-faq-q\">What does a self-hosted AI platform mean?<\/summary>\n            <p class=\"tx-faq-a\">A self-hosted AI platform runs in your own infrastructure\u2014either on-premises on your servers, in your private cloud, or air-gapped without an internet connection. All platform-side data such as prompts, documents, user permissions, and logs remains physically in-house. auraHub is designed as a self-hosted AI platform.  <\/p>\n          <\/details>\n        <\/div>\n        <div class=\"tx-faq-item\">\n          <details>\n            <summary class=\"tx-faq-q\">Which security levels does auraHub offer at the language model?<\/summary>\n            <p class=\"tx-faq-a\">Three levels. Level 1 uses a local language model with no external data outflow. Level 2 uses Azure OpenAI in an EU region with a data processing agreement via Microsoft. Level 3 uses the direct APIs of OpenAI or Anthropic, safeguarded via Standard Contractual Clauses. The level is chosen per use case.    <\/p>\n          <\/details>\n        <\/div>\n        <div class=\"tx-faq-item\">\n          <details>\n            <summary class=\"tx-faq-q\">What is Zero Data Retention?<\/summary>\n            <p class=\"tx-faq-a\">Zero Data Retention is a premium option for regulated data. Inputs and outputs are no longer stored after the response, are not written to abuse logs, and are not subjected to human review. Processing takes place purely in memory. Anthropic, Microsoft Azure, and OpenAI offer ZDR as part of enterprise contracts.   <\/p>\n          <\/details>\n        <\/div>\n        <div class=\"tx-faq-item\">\n          <details>\n            <summary class=\"tx-faq-q\">Is training done on our data?<\/summary>\n            <p class=\"tx-faq-a\">No For API and business plans, it is contractually assured that no training takes place on the transmitted data. This is a key difference from consumer apps such as ChatGPT Free, where data may be used for model training by default. <\/p>\n          <\/details>\n        <\/div>\n        <div class=\"tx-faq-item\">\n          <details>\n            <summary class=\"tx-faq-q\">How long are inputs stored outside the organization?<\/summary>\n            <p class=\"tx-faq-a\">Not at all with a local LLM. Azure OpenAI EU stores data by default for up to 30 days for abuse detection in an EU region. The OpenAI API stores data by default for up to 30 days; Anthropic for 7 days (reduced from 30 since September 2025). With Zero Data Retention, any storage after the response is eliminated.   <\/p>\n          <\/details>\n        <\/div>\n        <div class=\"tx-faq-item\">\n          <details>\n            <summary class=\"tx-faq-q\">Which GDPR building blocks are covered?<\/summary>\n            <p class=\"tx-faq-a\">Six building blocks: data processing agreement under Article 28; third-country transfer safeguarded via Standard Contractual Clauses (not applicable for local or EU); no training on your data; data minimization and short retention, with optional Zero Data Retention; open source and auditability of the platform source code; granular access control with roles, group permissions, and SSO via OpenID Connect.<\/p>\n          <\/details>\n        <\/div>\n        <div class=\"tx-faq-item\">\n          <details>\n            <summary class=\"tx-faq-q\">Which level fits which use case?<\/summary>\n            <p class=\"tx-faq-a\">Highly sensitive client or patient data runs on a local language model. Standard business data with an EU requirement runs via Azure OpenAI EU. Non-critical workloads run via the direct API of OpenAI or Anthropic. Where cloud models touch particularly sensitive data, Zero Data Retention safeguards processing without any storage.   <\/p>\n          <\/details>\n        <\/div>\n      <\/section>\n      <!-- AUTHOR -->\n      <div class=\"tx-author\">\n        <div class=\"tx-author-avatar\" aria-label=\"Initialen Oliver Range\">OR<\/div>\n        <div>\n          <div class=\"tx-author-handle\">@oliverrange<\/div>\n          <div class=\"tx-author-name\">Oliver Range<\/div>\n          <div class=\"tx-author-role\">Founder auraNexus.ai \u00b7 AI Manager (T\u00dcV)<\/div>\n          <p class=\"tx-author-bio\">Founder of several digital companies, including Die Medialysten (exit to Linkfluence). Consultant for GDPR-compliant AI platforms in the DACH mid-market. Develops AI applications with <a href=\"https:\/\/auranexus.ai\/en\/\">auraNexus.ai<\/a> for communications, tax advisory, law firms, healthcare, and manufacturing.  <\/p>\n        <\/div>\n      <\/div>\n      <!-- READ NEXT -->\n      <section class=\"tx-readnext\">\n        <div class=\"tx-readnext-label\">\/\/ read next<\/div>\n        <div class=\"tx-readnext-grid\">\n          <a href=\"https:\/\/auranexus.ai\/en\/aurahub\/\" class=\"tx-readnext-card\">\n            <div class=\"tx-readnext-cat\">auraHub<\/div>\n            <div class=\"tx-readnext-title\">AI platform without prompting, GDPR-compliant<\/div>\n          <\/a>\n          <a href=\"https:\/\/auranexus.ai\/en\/products\/\" class=\"tx-readnext-card\">\n            <div class=\"tx-readnext-cat\">Product overview<\/div>\n            <div class=\"tx-readnext-title\">Overview of AI apps for teams<\/div>\n          <\/a>\n          <a href=\"https:\/\/auranexus.ai\/en\/how-aurair-redefines-investor-relations\/\" class=\"tx-readnext-card\">\n            <div class=\"tx-readnext-cat\">auraIR<\/div>\n            <div class=\"tx-readnext-title\">AI for Investor Relations<\/div>\n          <\/a>\n          <a href=\"https:\/\/auranexus.ai\/en\/blog\/\" class=\"tx-readnext-card\">\n            <div class=\"tx-readnext-cat\">Blog<\/div>\n            <div class=\"tx-readnext-title\">All posts on AI strategy and practical tips<\/div>\n          <\/a>\n        <\/div>\n      <\/section>\n    <\/main>\n    <!-- MARGIN NOTES -->\n    <aside class=\"tx-margin\">\n      <div>\n        <div class=\"tx-note\">\n          <span class=\"tx-note-label\">\/\/ Core idea<\/span>\n          <div class=\"tx-note-quote\">The platform governs data storage. The model decides <em>everything else<\/em>. <\/div>\n        <\/div>\n        <div class=\"tx-note\">\n          <span class=\"tx-note-label\">\/\/ Layer 1<\/span>\n          <div class=\"tx-note-quote\">Prompts, documents, users, logs. <em>Everything stays<\/em> in-house.<\/div>\n        <\/div>\n        <div class=\"tx-note\">\n          <span class=\"tx-note-label\">\/\/ Level selection<\/span>\n          <span class=\"tx-note-stat\" data-count=\"3\">3<\/span>\n          <div class=\"tx-note-stat-label\">Security levels, configurable per use case. Not as a blanket setting. <\/div>\n        <\/div>\n        <div class=\"tx-note\">\n          <span class=\"tx-note-label\">\/\/ Anthropic retention<\/span>\n          <span class=\"tx-note-stat\" data-count=\"7\">7<\/span>\n          <div class=\"tx-note-stat-label\">Days of standard retention. Reduced from 30 days since September 2025. <\/div>\n        <\/div>\n        <div class=\"tx-note\">\n          <span class=\"tx-note-label\">\/\/ Premium<\/span>\n          <div class=\"tx-note-quote\">Zero Data Retention. <em>No storage<\/em> after the response.<\/div>\n        <\/div>\n        <div class=\"tx-note\">\n          <span class=\"tx-note-label\">\/\/ Note on Anthropic<\/span>\n          <div class=\"tx-note-quote\">Security classifiers are retained <em>even under ZDR<\/em>.<\/div>\n        <\/div>\n      <\/div>\n    <\/aside>\n  <\/div>\n<\/article>\n<script>\n(function() {\n  if (!('IntersectionObserver' in window) || !window.requestAnimationFrame) return;\n  var reduced = window.matchMedia && window.matchMedia('(prefers-reduced-motion: reduce)').matches;\n  var targets = document.querySelectorAll('.tx-inline-num[data-count], .tx-note-stat[data-count]');\n  if (!targets.length) return;\n  var observer = new IntersectionObserver(function(entries) {\n    entries.forEach(function(entry) {\n      if (!entry.isIntersecting || entry.target.dataset.counted) return;\n      var el = entry.target;\n      el.dataset.counted = '1';\n      el.classList.add('tx-num-revealed');\n      var target = parseInt(el.dataset.count, 10);\n      if (isNaN(target)) return;\n      if (reduced) { el.textContent = target; return; }\n      var duration = 1300;\n      var startTime = performance.now();\n      el.textContent = '0';\n      function step(now) {\n        var elapsed = now - startTime;\n        var progress = Math.min(elapsed \/ duration, 1);\n        var eased = 1 - Math.pow(1 - progress, 3);\n        el.textContent = Math.round(eased * target);\n        if (progress < 1) requestAnimationFrame(step);\n        else el.textContent = target;\n      }\n      requestAnimationFrame(step);\n    });\n  }, { threshold: 0.4, rootMargin: '0px 0px -10% 0px' });\n  targets.forEach(function(t) { observer.observe(t); });\n})();\n<\/script>\n\n<!--\n=============================================================\nMEDIA-SLOTS - auraHub DSGVO-Beitrag - 2026-05-28 (alle 4 befuellt)\n=============================================================\n\n-------------------------------------------------------------\nSLOT 1 - HERO (16:9) - BEFUELLT\n-------------------------------------------------------------\nPlatzierung: nach Sektion 01, oberhalb der tx-inline-numbers\nInhalt: Foto Arbeitsplatz mit auraHub-Dashboard und Blick auf den Koelner Dom\nURL: https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Moderner-Arbeitsplatz-mit-auraHub-Dashboard-und-Blick-auf-den-Koelner-Dom.png\nAlt-Text: Moderner Arbeitsplatz mit auraHub-Dashboard auf einem Monitor, im Hintergrund Blick auf den Koelner Dom.\nBildunterschrift: Arbeitsplatz mit auraHub-Dashboard und Blick auf den Koelner Dom. Die Plattform laeuft in Ihrem Haus, Sie sehen sie jeden Tag.\n\n-------------------------------------------------------------\nSLOT 2 - PLATTFORM-VISUALISIERUNG (16:9) - BEFUELLT\n-------------------------------------------------------------\nPlatzierung: innerhalb Sektion 02 Architektur, nach dem ersten Absatz (vor H3 \"Was bei Ihnen bleibt\")\nInhalt: Foto Server im Rechenzentrum mit moderner Netzwerk-Infrastruktur\nURL: https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Server-im-Rechenzentrum-mit-moderner-Netzwerk-Infrastruktur.png\nAlt-Text: Server im Rechenzentrum mit moderner Netzwerk-Infrastruktur. Symbolbild fuer die selbst gehostete Plattform-Ebene von auraHub.\nBildunterschrift: Die physische Plattform-Ebene: Server in Ihrer eigenen oder Ihrer Private-Cloud-Infrastruktur, auf denen auraHub direkt laeuft.\n\n-------------------------------------------------------------\nSLOT 3 - ARCHITEKTUR-DIAGRAMM (16:9) - BEFUELLT (SVG)\n-------------------------------------------------------------\nPlatzierung: innerhalb Sektion 02 Architektur, nach dem H3 \"Was das Sprachmodell sieht\"\nInhalt: Inline-SVG, Plattform-Modell-Trennung. Links Kundeninfrastruktur mit auraHub-Inhalten, rechts dunkle Box mit drei Sicherheitsstufen am Sprachmodell, dazwischen oranger Pfeil \"API-Aufruf nach gewaehlter Stufe\".\nRekonstruktion aus PDF-Seite 1 (\"Ebene Plattform\").\nAspect: 16:9, viewBox 1600x900\nAlt-Text: Architektur-Diagramm: auraHub-Plattform in der Kundeninfrastruktur links, drei Sicherheitsstufen am Sprachmodell rechts, verbunden ueber einen API-Pfeil.\nBildunterschrift: Plattform und Sprachmodell sind getrennt. Was bei Ihnen bleibt, und was die Wahl der Stufe entscheidet.\n\n-------------------------------------------------------------\nSLOT 4 - DREI SICHERHEITSSTUFEN (16:9) - BEFUELLT (SVG)\n-------------------------------------------------------------\nPlatzierung: innerhalb Sektion 03 Drei Stufen, am Ende nach den Begriffsklarheit-Erlaeuterungen\nInhalt: Inline-SVG, drei Spalten nebeneinander. Spalte 1 (dunkel) Lokales LLM mit vLLM, LM Studio, Jan.ai. Spalte 2 (hell mit Teal-Border) Azure OpenAI EU mit Sweden, France. Spalte 3 (hell, neutral) OpenAI\/Anthropic direkt. Jede Spalte mit Eyebrow \"STUFE 1\/2\/3\", Stufenname, Modellname, drei Eigenschafts-Bloecke und Eignungs-Footer.\nRekonstruktion aus PDF-Seite 2 (\"Drei Sicherheitsstufen zur Wahl\").\nAspect: 16:9, viewBox 1800x1000\nAlt-Text: Drei Sicherheitsstufen fuer das Sprachmodell: Lokales LLM, Azure OpenAI EU und OpenAI oder Anthropic direkt, jeweils mit Verarbeitungsort, Speicherung und Eignung.\nBildunterschrift: Drei Sicherheitsstufen am Sprachmodell. Lokal fuer hochsensible Daten, EU fuer regulaere Geschaeftsdaten, SCC-abgesichert fuer unkritische Workloads.\n\n=============================================================\n-->\t\t\t\t<\/div>\n\t\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t\t\t<\/div>\n\t\t","protected":false},"excerpt":{"rendered":"<p>auraHub \u00b7 GDPR and language models The platform is not the problem. If auraHub runs on your own infrastructure, your data stays with you. Who decides what happens to prompts and responses is determined by your choice of language model. Three levels for three use cases. Oliver Range auraNexus.ai \u00b7 May 28, 2026 9 min [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":3935,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_angie_page":false,"page_builder":"","footnotes":""},"categories":[51],"tags":[652,235,653,656,240,651,655,654],"class_list":["post-3936","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-tools","tag-ai-compliance","tag-aurahub","tag-enterprise-ai-platform","tag-gdpr-ai-platform","tag-on-premises-ai","tag-private-cloud-ai","tag-secure-ai-integration","tag-three-security-levels"],"yoast_head":"<!-- This site is optimized with the Yoast SEO Premium plugin v28.0 (Yoast SEO v28.0) - https:\/\/yoast.com\/product\/yoast-seo-premium-wordpress\/ -->\n<title>The platform is not the problem |<\/title>\n<meta name=\"description\" content=\"GDPR-compliant AI platform: platform and language model separated. auraHub runs on-premises, in a private cloud, or air-gapped. Two layers, three security levels. Learn more.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"The platform is not the problem\" \/>\n<meta property=\"og:description\" content=\"GDPR-compliant AI platform: platform and language model separated. auraHub runs on-premises, in a private cloud, or air-gapped. Two layers, three security levels. Learn more.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/\" \/>\n<meta property=\"article:published_time\" content=\"2026-05-28T16:20:48+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2026-06-25T07:17:17+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp\" \/>\n\t<meta property=\"og:image:width\" content=\"1920\" \/>\n\t<meta property=\"og:image:height\" content=\"1281\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/webp\" \/>\n<meta name=\"author\" content=\"O.Range\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data1\" content=\"O.Range\" \/>\n\t<meta name=\"twitter:label2\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data2\" content=\"12 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/\"},\"author\":{\"name\":\"O.Range\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#\\\/schema\\\/person\\\/3d27829ac34c5bbd905e797bc9f2f481\"},\"headline\":\"The platform is not the problem\",\"datePublished\":\"2026-05-28T16:20:48+00:00\",\"dateModified\":\"2026-06-25T07:17:17+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/\"},\"wordCount\":2465,\"publisher\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/auranexus.ai\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp\",\"keywords\":[\"AI compliance\",\"auraHub\",\"Enterprise AI platform\",\"GDPR AI platform\",\"On-Premises AI\",\"Private cloud AI\",\"Secure AI integration\",\"Three security levels\"],\"articleSection\":[\"AI tools\"],\"inLanguage\":\"en-US\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/\",\"url\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/\",\"name\":\"The platform is not the problem |\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/auranexus.ai\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp\",\"datePublished\":\"2026-05-28T16:20:48+00:00\",\"dateModified\":\"2026-06-25T07:17:17+00:00\",\"description\":\"GDPR-compliant AI platform: platform and language model separated. auraHub runs on-premises, in a private cloud, or air-gapped. Two layers, three security levels. Learn more.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/#breadcrumb\"},\"inLanguage\":\"en-US\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/#primaryimage\",\"url\":\"https:\\\/\\\/auranexus.ai\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp\",\"contentUrl\":\"https:\\\/\\\/auranexus.ai\\\/wp-content\\\/uploads\\\/2026\\\/05\\\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp\",\"width\":1920,\"height\":1281,\"caption\":\"Employees working together on tablets and laptops in a modern meeting room with bright architecture and glass frontage\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/the-platform-is-not-the-problem\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/ai-consulting-for-companies-auranexus-ai\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"The platform is not the problem\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#website\",\"url\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/\",\"name\":\"auranexus.ai\",\"description\":\"\",\"publisher\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en-US\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#organization\",\"name\":\"auranexus.ai\",\"url\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/auranexus.ai\\\/wp-content\\\/uploads\\\/2026\\\/01\\\/auraNexus_App_Button.svg\",\"contentUrl\":\"https:\\\/\\\/auranexus.ai\\\/wp-content\\\/uploads\\\/2026\\\/01\\\/auraNexus_App_Button.svg\",\"width\":156,\"height\":156,\"caption\":\"auranexus.ai\"},\"image\":{\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"http:\\\/\\\/linkedin.com\\\/company\\\/auranexus-ai\",\"https:\\\/\\\/www.instagram.com\\\/auranexus.ai\\\/\"],\"description\":\"KI-Beratung f\u00fcr Unternehmen. Schulungen, Custom GPTs und Automatisierung f\u00fcr den deutschen Mittelstand.\",\"email\":\"info@auranexus.ai\",\"telephone\":\"+49 (0) 2234 4351436\",\"legalName\":\"auraNexus.ai\",\"foundingDate\":\"2025-12-10\",\"vatID\":\"DE459170091\",\"numberOfEmployees\":{\"@type\":\"QuantitativeValue\",\"minValue\":\"1\",\"maxValue\":\"10\"}},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/#\\\/schema\\\/person\\\/3d27829ac34c5bbd905e797bc9f2f481\",\"name\":\"O.Range\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/560469edd5817a20f96d3053888137f72e76ab7714ed8c16e06b506f353c3ac6?s=96&d=mm&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/560469edd5817a20f96d3053888137f72e76ab7714ed8c16e06b506f353c3ac6?s=96&d=mm&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/560469edd5817a20f96d3053888137f72e76ab7714ed8c16e06b506f353c3ac6?s=96&d=mm&r=g\",\"caption\":\"O.Range\"},\"sameAs\":[\"https:\\\/\\\/auranexus.ai\"],\"url\":\"https:\\\/\\\/auranexus.ai\\\/en\\\/author\\\/sdfkxjhnxz\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO Premium plugin. -->","yoast_head_json":{"title":"The platform is not the problem |","description":"GDPR-compliant AI platform: platform and language model separated. auraHub runs on-premises, in a private cloud, or air-gapped. Two layers, three security levels. Learn more.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/","og_locale":"en_US","og_type":"article","og_title":"The platform is not the problem","og_description":"GDPR-compliant AI platform: platform and language model separated. auraHub runs on-premises, in a private cloud, or air-gapped. Two layers, three security levels. Learn more.","og_url":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/","article_published_time":"2026-05-28T16:20:48+00:00","article_modified_time":"2026-06-25T07:17:17+00:00","og_image":[{"width":1920,"height":1281,"url":"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp","type":"image\/webp"}],"author":"O.Range","twitter_card":"summary_large_image","twitter_misc":{"Written by":"O.Range","Est. reading time":"12 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/#article","isPartOf":{"@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/"},"author":{"name":"O.Range","@id":"https:\/\/auranexus.ai\/en\/#\/schema\/person\/3d27829ac34c5bbd905e797bc9f2f481"},"headline":"The platform is not the problem","datePublished":"2026-05-28T16:20:48+00:00","dateModified":"2026-06-25T07:17:17+00:00","mainEntityOfPage":{"@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/"},"wordCount":2465,"publisher":{"@id":"https:\/\/auranexus.ai\/en\/#organization"},"image":{"@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/#primaryimage"},"thumbnailUrl":"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp","keywords":["AI compliance","auraHub","Enterprise AI platform","GDPR AI platform","On-Premises AI","Private cloud AI","Secure AI integration","Three security levels"],"articleSection":["AI tools"],"inLanguage":"en-US"},{"@type":"WebPage","@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/","url":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/","name":"The platform is not the problem |","isPartOf":{"@id":"https:\/\/auranexus.ai\/en\/#website"},"primaryImageOfPage":{"@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/#primaryimage"},"image":{"@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/#primaryimage"},"thumbnailUrl":"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp","datePublished":"2026-05-28T16:20:48+00:00","dateModified":"2026-06-25T07:17:17+00:00","description":"GDPR-compliant AI platform: platform and language model separated. auraHub runs on-premises, in a private cloud, or air-gapped. Two layers, three security levels. Learn more.","breadcrumb":{"@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/#breadcrumb"},"inLanguage":"en-US","potentialAction":[{"@type":"ReadAction","target":["https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/"]}]},{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/#primaryimage","url":"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp","contentUrl":"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/05\/Modernes-Teammeeting-in-offenem-Buero-mit-digitaler-Zusammenarbeit-scaled.webp","width":1920,"height":1281,"caption":"Employees working together on tablets and laptops in a modern meeting room with bright architecture and glass frontage"},{"@type":"BreadcrumbList","@id":"https:\/\/auranexus.ai\/en\/the-platform-is-not-the-problem\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/auranexus.ai\/en\/ai-consulting-for-companies-auranexus-ai\/"},{"@type":"ListItem","position":2,"name":"The platform is not the problem"}]},{"@type":"WebSite","@id":"https:\/\/auranexus.ai\/en\/#website","url":"https:\/\/auranexus.ai\/en\/","name":"auranexus.ai","description":"","publisher":{"@id":"https:\/\/auranexus.ai\/en\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/auranexus.ai\/en\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en-US"},{"@type":"Organization","@id":"https:\/\/auranexus.ai\/en\/#organization","name":"auranexus.ai","url":"https:\/\/auranexus.ai\/en\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/auranexus.ai\/en\/#\/schema\/logo\/image\/","url":"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/01\/auraNexus_App_Button.svg","contentUrl":"https:\/\/auranexus.ai\/wp-content\/uploads\/2026\/01\/auraNexus_App_Button.svg","width":156,"height":156,"caption":"auranexus.ai"},"image":{"@id":"https:\/\/auranexus.ai\/en\/#\/schema\/logo\/image\/"},"sameAs":["http:\/\/linkedin.com\/company\/auranexus-ai","https:\/\/www.instagram.com\/auranexus.ai\/"],"description":"KI-Beratung f\u00fcr Unternehmen. Schulungen, Custom GPTs und Automatisierung f\u00fcr den deutschen Mittelstand.","email":"info@auranexus.ai","telephone":"+49 (0) 2234 4351436","legalName":"auraNexus.ai","foundingDate":"2025-12-10","vatID":"DE459170091","numberOfEmployees":{"@type":"QuantitativeValue","minValue":"1","maxValue":"10"}},{"@type":"Person","@id":"https:\/\/auranexus.ai\/en\/#\/schema\/person\/3d27829ac34c5bbd905e797bc9f2f481","name":"O.Range","image":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/secure.gravatar.com\/avatar\/560469edd5817a20f96d3053888137f72e76ab7714ed8c16e06b506f353c3ac6?s=96&d=mm&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/560469edd5817a20f96d3053888137f72e76ab7714ed8c16e06b506f353c3ac6?s=96&d=mm&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/560469edd5817a20f96d3053888137f72e76ab7714ed8c16e06b506f353c3ac6?s=96&d=mm&r=g","caption":"O.Range"},"sameAs":["https:\/\/auranexus.ai"],"url":"https:\/\/auranexus.ai\/en\/author\/sdfkxjhnxz\/"}]}},"_links":{"self":[{"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/posts\/3936","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/comments?post=3936"}],"version-history":[{"count":1,"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/posts\/3936\/revisions"}],"predecessor-version":[{"id":3937,"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/posts\/3936\/revisions\/3937"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/media\/3935"}],"wp:attachment":[{"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/media?parent=3936"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/categories?post=3936"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/auranexus.ai\/en\/wp-json\/wp\/v2\/tags?post=3936"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}