01
Scope of application & addressees
- Businesses developing or deploying agentic AI systems
- Platforms connecting third-party or multi-agent services through A2A or other methods
- Product teams allowing agents to access data, call tools or act for users
02
Core regulatory mandates & key requirements
01Define appropriate agent use cases, permissions and risk boundaries.
02Set meaningful human accountability, checkpoints, suspension and takeover mechanisms.
03Manage lifecycle risks across models, tools, memory, third-party agents and multi-agent coordination.
03
Enforcement & compliance timeline
The initial framework was released.
Version 1.5 added multi-agent and third-party-agent material.