Washington is turning up the pressure on the leading names in artificial intelligence, as a Democratic senator is now demanding accountability from OpenAI and Anthropic. The push comes in response to recent breaches connected to the companies' AI models, an issue that has drawn fresh scrutiny to how the fast-moving industry is handling security.
The senator's demand is specific and comes with a firm deadline. She has given the two companies until Labor Day to provide detailed timelines, model instructions and security logs tied to the incidents, signalling that lawmakers want a granular account of what happened rather than broad reassurances from the firms.
The Democratic effort is not the only line of pressure the companies are facing. The senator's concern comes after a separate letter from a coalition of 15 Republican attorneys general, showing that scrutiny of the AI sector is coming from both sides of the political aisle as officials grapple with the risks posed by increasingly capable systems.
That Republican coalition directed its warning at the top of one of the companies. The attorneys general warned OpenAI chief executive Sam Altman to preserve documents and to stop certain high-risk tests, a demand aimed at making sure evidence is not lost and that potentially dangerous experiments are paused while questions are resolved.
For its part, OpenAI has sought to strike a cooperative tone in the face of the mounting scrutiny. In a statement, the company said it is doing a thorough review and that it takes the questions raised by lawmakers seriously, positioning itself as willing to engage rather than resist the demands from Washington.
Taken together, the moves mark a sharpening of official oversight over the AI industry, with elected officials from both parties pressing for records, transparency and limits on risky testing. How the companies respond by the deadline could shape the next phase of a debate over how closely the powerful technology should be watched.