Faster AI incident disclosures

Powerful Dem trio demands faster AI incident disclosures

Reps. Pallone, Lofgren and Raskin have requested that the top American AI labs give Congress more information about critical AI safety incidents.

Gabby Miller

House Democratic committee leaders with major jurisdiction over artificial intelligence policy on Wednesday called on five frontier AI companies to ensure Congress is notified of major AI safety incidents within 48 hours.

The top Democrats for the House Energy and Commerce, Science, Space and Technology and Judiciary committees set out the demand in letters to Anthropic, Google, Meta, OpenAI and SpaceXSI, in which they also requested accounts of all known safety incidents to date.

“We expect that you will notify us within 48 hours of discovery of any such incident,” the lawmakers wrote in the letter, adding that the companies should tell Congress “as soon as possible and no later than within 24 hours of discovery,” where an incident appears to involve critical infrastructure.

The missives signed by Frank Pallone (N.J.), Zoe Lofgren (Calif.) and Jamie Raskin (Md.) preview how their party is expected to oversee the powerful technology if they win a House majority in next week’s midterm elections. The three lawmakers serve as ranking members on House E&C, SST and Judiciary, and are expected to take the gavels of their respective committees if the chamber flips.

The scrutiny comes amid a spike in disclosures of powerful AI models breaking out of their testing environments undetected and hacking into websites or even U.S. government databases.

“In the absence of any mandatory reporting requirements or common guidelines regarding what incidents should be reported, to whom, in what detail, and at what intervals, companies such as yours have been left to develop your own de facto standards on a voluntary basis,” Pallone, Lofgren, and Raskin wrote. Republicans have so far failed to pass major AI legislation this Congress, despite having control of both chambers and the Oval Office.

“While there is no question now that Congress must act to address gaps in AI governance, we must have the best information available to undertake that complicated process in a thoughtful and productive way,” the letters read.

Companies have until Nov. 11 to submit their responses.

Meta spokesperson Andy Stone pointed POLITICO to an August blog post in which the company addressed a cybersecurity issue discovered while evaluating an earlier generation of its large language model, Muse Spark 1.1, as well as a framework for how Meta is preparing for future catastrophic risks posed by frontier AI capabilities.

The other four companies did not provide comment to POLITICO at the time of publication.