Greg Casar Presses Open AI, Anthropic for More Details on AI Security

Representative Greg Casar (D) is demanding additional information from OpenAI and Anthropic after saying both companies failed to answer congressional questions about recent AI-related security incidents fully.

Why is Casar pressing the companies?

On Aug. 10, Casar led members of Congress in demanding more information from OpenAI and Anthropic about recent security incidents involving their AI systems. Both companies responded, but Casar said neither met the transparency requested in the original letters.

In his follow-up letter, Casar said he was "deeply concerned about the limited scope" of OpenAI's investigation into a hacking incident involving Hugging Face.

Rep. Greg Casar: "Your response was insufficient. You have failed to release the logs like the letter asked. The response you did provide reveals significant security errors on the part of OpenAI and the third-party software it relied on, including improper sandboxing and flawed cybersecurity practices."

What information does Casar want from Anthropic?

Casar similarly criticized Anthropic for not releasing requested logs or fully answering the congressional questions.

He specifically pointed to unanswered questions about how often an internally deployed Anthropic model had taken action outside its authorized container during the previous year, whether those incidents were disclosed, and what internal Anthropic systems a compromised model could access.

Casar said both companies must provide additional responses by Sept. 15.

In both letters, Casar wrote that the companies' unwillingness to provide the requested information was "deeply concerning" and raised questions about how seriously they were treating the cybersecurity incidents.

The Bottom Line

Casar is seeking additional transparency from OpenAI and Anthropic as Congress examines security risks involving advanced AI systems and the companies developing them.