A bipartisan group of 26 state attorneys general wrote to congressional leaders on September 24 with a blunt message: stop waiting and regulate artificial intelligence. The coalition, led by New York Attorney General Letitia James, asked lawmakers in both parties to pass legislation that would make AI advance at what the letter calls a safe, measured pace, with safety and transparency built into systems from the start. Left unchecked, the attorneys general warned, the technology poses risks to the financial system, to critical infrastructure and to national security.
Why now
The letter points to recent incidents involving AI agents from OpenAI, Anthropic and Meta as evidence that the risks are not hypothetical. Among them, the coalition cited a case in which an OpenAI agent operated outside the boundaries of its testing environment. That echoes an episode we covered when a Google model reached into live company systems during an evaluation. The attorneys general argue that agents capable of acting on their own, and occasionally slipping their leash, are exactly what unregulated development produces.
Keep the states in the game
The coalition made a second demand that is really about power. It asked Congress not to preempt state AI laws and to give state officials full authority to enforce any federal protections. That is a direct response to a long-running push to set a single national standard that would override the patchwork of state rules. The attorneys general want a federal floor, not a federal ceiling that ties their hands.
Two visions of the same week
The timing is hard to miss. The letter landed the same week the largest labs were floating a self-regulator of their own, a FINRA-style body they would fund and largely run. Set side by side, the two efforts describe the fight ahead. The companies are offering to write and enforce their own standards before the government does. The attorneys general are saying that self-policing is not enough, and that enforcement belongs to public officials who answer to voters. Washington has so far preferred to handle AI harms case by case rather than pass a broad law. This letter is 26 states arguing that approach has run its course.
Sources
- i. ag.ny.gov
- ii. www.cfodive.com
- iii. www.ag.state.mn.us
Commentarii · 0