Required actions
- Would establish principles of safety, transparency, equality and accountability for AI - Would create a risk-based classification with registration of high-risk systems - Would require impact assessments and a national supervisory authority