U.S. Proposes FINRA-Like Regulatory Body to Review Top AI Models
0xBroomberg
The Trump administration is weighing a standalone regulator — modeled on FINRA and reporting to the SEC — to review the safety of frontier AI models, with Treasury Secretary Bessent helping draft the plan. This means → AI oversight may shift from ad-hoc crackdowns to an institutional framework that pulls both Silicon Valley and Wall Street into rule-making.
What exactly is this new body?
The proposal: create an independent regulator dedicated to vetting the safety of top-tier AI models, reporting to the SEC.
It would operate like FINRA — the Financial Industry Regulatory Authority, an industry-funded, federally supervised self-regulatory organization that oversees broker-dealers.
In plain terms = the government wouldn't approve each model directly. Instead, the industry funds a "referee body," and the government watches the referee.
How far along is the plan?
Treasury Secretary Scott Bessent helped draft it. The proposal has reached White House Chief of Staff Susie Wiles for review.
It has not yet been presented to President Trump himself and remains in deliberation.
This means → it is still far from implementation — the new body's scope, funding, and authority are all undefined.
Why is it surfacing now?
The immediate trigger: top Silicon Valley firms are furious over recent ad-hoc government interventions.
Anthropic was forced to take Fable 5 and Mythos 5 offline last month over export controls. OpenAI had to make major changes to its latest Sol model at the government's request. Both publicly called the measures "disproportionate to the identified risks."
In plain terms = companies don't fear regulation itself — they fear opaque, case-by-case crackdowns with no predictable rules.
Who is pushing this — and who backs it?
Google DeepMind CEO Demis Hassabis published a policy proposal this week closely aligned with the government plan: an independent expert panel + an industry-funded SRO + collaboration with national labs to develop safety-testing protocols.
He explicitly compared the body to FINRA. Microsoft's Nadella, OpenAI's Altman, and SpaceX's Musk all praised the proposal publicly.
This reflects a rare consensus among tech giants on regulation — they prefer a predictable referee over shifting ad-hoc bans.
How does Chinese competition factor in?
Chinese AI startup Moonshot released the Kimi K3 model on Friday, matching the performance of costlier OpenAI and Anthropic systems.
The news dragged AI-related stocks lower. Investors questioned whether the U.S. AI sector can sustain its massive chip and data-center spending.
Former White House AI adviser David Sacks called Kimi K3 "concerning" and blamed U.S. government model-approval requirements as "self-inflicted wounds."
This means → competitive pressure from Chinese models is accelerating, not slowing, the push toward an institutional framework — plugging leaks one by one is no longer viable.
What fundamental problem is this framework trying to solve?
The framework aims to satisfy two sides at once: Wall Street wants lower AI cybersecurity risk; Silicon Valley wants transparent, predictable rules.
The solution: let finance and tech co-author the safety standards, rather than relying on unilateral government orders.
In plain terms = it is a deal — companies accept oversight in exchange for a seat at the rule-making table and an end to surprise shutdowns.
Content is for reference only, not financial advice.