ACAAI (not the Brazilian fruit🍇) Accountability by Design for Agentic AI: A Lifecycle Framework for Engineering Accountability in Autonomous AI Systems by @IsabBarb These are its 6 accountability control domains: 1. Organizational governance 2. Human oversight, consent, and decision authority 3. Identity, authority, and data governance 4. Observability, explainability, and evidence 5. Runtime safety and assurance 6. Incident response and recovery This is immediately practical if you (or your organization) are designing, building, deploying, overseeing, or regulating agentic AI systems. It gives a clear, ready-to-use map of the areas that must be covered so nothing critical is overlooked. You can use the 6 domains as a checklist or architecture review guide, mapping your existing policies, technical safeguards, logging, identity management, and incident plans against them. It helps satisfy emerging regulatory expectations (human oversight, transparency, risk management, auditability) without inventing everything from scratch. Even if you are not a technical expert, the structure makes it easier to ask the right questions of vendors or internal developers: “How is identity and authority handled?” “What runtime safety controls exist?” “Can we reconstruct any decision with evidence?”, etc.