US Weighs FINRA-Style AI Regulator — Industry Self-Inspection Model Is the Structural Opposite of the EU AI Act; Trans-Atlantic Compliance Now Requires Two-Clock Planning

A plan taking shape in Washington would create an independent AI regulator modeled on FINRA — a private, industry-funded organization reporting to the SEC, staffed largely from the industry it regulates — to vet the most capable AI models before release. Treasury Secretary Scott Bessent helped develop the proposal; it now sits with White House chief of staff Susie Wiles. In the first phase, labs would voluntarily submit frontier models for up to 30 days of review targeting dangerous capabilities (autonomous action, guardrail bypass, deception, offensive cyber ability, biological or nuclear risk); passing the assessment would eventually become a condition of deploying a model in the US market. Nothing is law yet — the voluntary first phase means labs could opt in before any mandate exists. The EU contrast is explicit and material: the EU AI Act is enforced by public authorities with real financial penalties from August 2, and no industry-funded body sits between a lab and the regulator. For law firms and legal departments deploying AI systems on both sides of the Atlantic, the governance gap produces direct compliance planning complexity: the same model may clear a US industry review weeks before it satisfies EU Article 50 transparency obligations; procurement and compliance workflows must be built for divergence rather than convergence; and the two regimes' incompatible assumptions about who should set AI rules — industry self-regulation versus public enforcement — are now the terrain on which trans-Atlantic AI legal advice will be delivered for the next several years.