◐ EU AI Act
How we comply.
AURA was designed against the EU AI Act before the Act was finalised. Here's what we do — and what stays your right.
Disclosure of AI-generated content
Art. 50 EU AI Act requires deployers of AI systems generating synthetic content to disclose this clearly. Every avatar profile, every generated post, every synthetic audio carries machine-readable metadata identifying it as AI. Where the destination platform supports human-visible labelling (TikTok, Instagram, YouTube), we apply it automatically.
Watermarking
AI-generated images and audio carry invisible watermarks (C2PA-compatible). Video clips include visible "AI-generated" disclosure in the first frame on platforms that don't render the platform-native label.
What is not disclosed
The identity of the human behind the avatar. This is your fundamental data protection right under Art. 8 GRCh and Art. 7–9 DSGVO. The AI Act does not require and does not authorise disclosing the natural person controlling an AI persona — and AURA's trustee architecture forbids it.
High-risk areas we don't enter
AURA avatars cannot be used for political campaigning, election interference, medical or legal advice, employment decisions, or impersonation of real living public figures. These uses are blocked at the system-prompt level and audited.
Subliminal & manipulative content
Per Art. 5 AI Act: avatars may not exploit cognitive vulnerabilities, age-related vulnerabilities, or use subliminal techniques. Our generation pipeline includes safeguards against these patterns.
Your right to know
You can request a complete export of every piece of training data, every system prompt, and every model decision applied to your avatar at any time. Within 14 days you receive a machine-readable archive.
For specific compliance questions: legal@aura.beraterag.com