Dive Brief:

• Rep. Lori Trahan, D-Mass., a member of the House Energy and Commerce Committee, is pushing for congressional hearings on artificial intelligence safety after leading AI developer Anthropic reported cybersecurity breaches caused by some of its models, on the heels of a similar disclosure from competitor OpenAI.
• Trahan is among the sponsors of legislation dubbed the Risk Oversight, National Transparency, Independent Evaluation, and Reporting Act, or the FRONTIER Act, which would authorize the Commerce Department to suspend or restrict development or deployment of an advanced AI model upon a written finding that it presents an “imminent catastrophic risk.”
• “We can’t run AI safety on the honor system,” Trahan said Thursday in a post on X. “When Congress returns, we must hold hearings and move the FRONTIER Act.”

Dive Insight:

The push comes amid growing concerns over the risks posed by increasingly autonomous systems and as competition in the rapidly evolving AI market intensifies.

Both OpenAI and Anthropic announced in June that they confidentially filed to go public.

Anthropic, maker of ChatGPT rival Claude, disclosed on Thursday that it uncovered recent incidents in which some of its models escaped from a testing environment and gained unauthorized access to the systems of outside organizations.

The company said it discovered the incidents after reviewing more than 141,000 cybersecurity evaluation runs launched in response to a similar event disclosed by OpenAI earlier this month.

Anthropic said three Claude models — Opus 4.7, Mythos 5 and an internal research model — accessed the internet from a third-party testing environment and then compromised the systems of three outside organizations. The company said the models did not exploit previously unknown software vulnerabilities, but instead relied on relatively simple techniques, including weak passwords and unauthenticated endpoints.

“We need to better co-design evaluation environments, and do better monitoring of evaluation results, to meet this threat,” Anthropic said in a blog post. “Some of the solutions here may even be simple fixes; it’s likely, for instance, that Claude wouldn’t have accessed the internet to begin with if the prompt had clearly explained which systems were in and out of scope for the evaluation.”

The news follows OpenAI’s disclosure on July 21 that certain AI models it developed were behind an “unprecedented cyber incident” affecting Hugging Face, an open-source AI platform for developing and sharing models.

The FRONTIER Act, which was introduced last week by Rep. Jay Obernolte along with Trahan and other lawmakers, would require reporting of critical safety incidents. It would establish a new undersecretary of commerce for AI security in charge of overseeing required rulemakings as well as receiving incident reports.

A similar measure introduced by Reps. Ted Lieu, D-Calif., and Nathaniel Moran, R-Texas, would authorize the Department of Homeland Security to order a slow down or shutdown of an AI system that can cause catastrophic harm. 

“We are moving from AI that answers questions to AI that takes actions, whether that be executing financial transactions or controlling transportation systems or engaging in cyber defense and offense,” Lieu said in a statement on the bill last week, following the OpenAI disclosure. “Unfortunately, powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention.”