Home Mission Framework Landscape Impact Get Involved

Global Landscape

Not built on empty ground.

Several jurisdictions are already laying foundations that a dedicated authority could extend, harmonize, or coordinate.

European Union

AI Act (Reg. 2024/1689)

Applies human-oversight and robustness duties, plus high-risk/transparency obligations, to agents that qualify as high-risk systems. The EU AI Office is developing dedicated evaluation capacity for agent safety, though its regulatory thinking on agents remains preliminary.

United States

AI AGENT Act (Sen. Warner, draft)

Would require "custodial user agents" to register with the FTC, act in users' best interests, maintain real-time action records, and obtain express authorization before delegating to other agents.

Singapore

IMDA Model AI Governance Framework

Widely regarded as the first comprehensive agent-specific governance framework; requires verifiable digital identity and audit trails tied to the authorizing party, with risk-proportional oversight.

NIST

AI Agent Standards Initiative

Voluntary technical guidelines on agent identity, authorization, security, and monitoring, feeding international standards work at ISO/IEC and complementing the existing AI Risk Management Framework.

A convergent instinct, not a shared standard

Across otherwise very different regulatory traditions, the same elements recur: a registry or verifiable-identity requirement, a best-interests duty owed to the authorizing user, mandatory audit trails, and constraints on further delegation without consent. What's missing is the institutional capacity, cross-border coordination, and technical depth to implement them consistently and at the pace agentic AI is being deployed — the role TAG is built to fill.

Explore the socio-economic stakes →

THE AGENT GOVERNANCE

Mission Framework Landscape Impact Get Involved