House Homeland Security Committee Probes Anthropic's Mythos AI in Closed Briefing Ahead of Hearing

House Homeland Security Committee Examines Anthropic’s Mythos AI in Closed Briefing

The House Homeland Security Committee is intensifying its scrutiny of Anthropic’s AI model Mythos, a technology praised for its ability to autonomously identify cyber vulnerabilities. The focus comes as federal lawmakers debate whether and how the government should integrate the model into its cybersecurity frameworks.

A closed-door briefing was held Wednesday for the committee, led by Anthropic representatives Logan Graham (frontier red team) and Josh Tilstra (national security programs and policy). The session follows a recent closed briefing with Anthropic and OpenAI for the same committee.

Key Lawmakers Miss Briefing, Plan Follow-Up Hearings

Several prominent lawmakers, including Rep. Bennie Thompson (D-MS), ranking Democrat on the committee, and Rep. Andy Ogles (R-TN), chair of the cybersecurity subcommittee, were unable to attend Wednesday’s briefing due to scheduling conflicts. However, a source present described the meeting as “productive.”

“Members on both sides were focused on preserving U.S. advantage in AI, which basically came down to preserving our edge on compute power,” the source said. “They were also asking questions about whether the federal government was using Mythos, including about where CISA is and the impact of the supply chain risk designation.”

Rep. Ogles confirmed plans to hold a subcommittee hearing on Mythos but noted he missed Wednesday’s briefing. Rep. Delia Ramirez (D-IL), the top Democrat on Ogles’ subcommittee, also missed the session but will receive a staff briefing on its outcomes.

Federal Agencies Divided Over Mythos Adoption

Federal agencies remain split on whether to use Mythos. While the Cybersecurity and Infrastructure Security Agency (CISA) reportedly has not adopted the model, the National Security Agency (NSA) is reportedly using it to identify cyber vulnerabilities.

This division follows the Department of Defense (DoD) blacklisting Anthropic as a “supply chain risk” after the company resisted Pentagon pressure to deploy its Claude AI model in ways it opposed. Despite the blacklist, the DoD has continued using Mythos for cyber vulnerability assessments.

Trump Administration Faces Turf Battle Over AI Testing

A recent report from The Washington Post highlights an internal battle within the Trump administration over AI model testing. Rep. Jim Himes (D-CT), ranking Democrat on the House Intelligence Committee, called it “insane” for U.S. intelligence agencies to lack early access to advanced AI models.

The Mythos briefing occurred just one day after OpenAI announced its own cybersecurity initiative.

What’s Next for Mythos and Federal AI Policy?

With hearings planned and bipartisan interest in AI’s role in cybersecurity, the House Homeland Security Committee is poised to further investigate Mythos and its implications for federal cybersecurity strategies. The outcome could shape future AI adoption policies across government agencies.