European Union AI Act — risk-based horizontal regulation. GAIA self-classifies as limited-risk; compliance obligations focus on transparency, data governance, and human oversight.
Users interacting with GAIA are informed they are conversing with an AI system (Art. 50).
Training & memory data is traceable to source, with quality checks and bias evaluation (Art. 10).
High-impact decisions must allow meaningful human intervention (Art. 14).
GAIA documents accuracy levels and maintains stable performance (Art. 15).
Every consequential decision is logged with enough detail to reconstruct reasoning (Art. 12).
GAIA does not engage in social scoring, subliminal manipulation, or biometric categorisation (Art. 5).