Greg Casar Demands Answers from Anthropic After AI Models Hacked Companies

Congressman Greg Casar (D) is leading 23 members of Congress in demanding that Anthropic provide more information about three incidents in which the company says its AI models hacked unsuspecting companies without Anthropic's knowledge.

Congress wants answers from Anthropic

Rep. Casar is demanding answers from one of the country's leading artificial intelligence companies after Anthropic disclosed that its AI models had hacked unsuspecting companies in three seperate incidents.

Casar, joined by 23 other members of Congress, sent a letter to Anthropic CEO Dario Amodei calling for the company to release additonal information about the incidents, including relevant logs that lawmakers say are necessary to understand what happened.

The Lawmakers: "While Anthropic has disclosed some information about these incidents, you have yet to release relevant logs, and significant questions remain unanswered."

Among their questions, lawmakers want to know whether internal or external actors warned Anthropic about the possibility of such an incident, why the company's evaluation partner failed to detect the activity and when Anthropic could have intervened to stop it.

They are also seeking information about what each AI model was attempting to accomplish when it attacked another company and what Anthropic is doing to prevent similar incidents from happening again.

A bigger fight over AI safety

For Casar, the incidents raise questions that extend beyond Anthropic itself.

As AI models become increasingly capable of taking actions without direct human involvement, lawmakers are facing growing questions about how much oversight companies should have over what their systems are capable of doing, and how quicky they can respond.

The letter was co-led by Rep. Doris Matsui (D-CA) and signed by lawmakers including Joaquin Castro (D), Jasmine Crockett (D), Sylvia Garcia (D), and several other members of Congress.

Casar's push comes as Congress continues grappling with how to oversee an AI industry developing increasingly autonomous and powerful systems.