Key Points
- Anthropic CEO Dario Amodei has scheduled a Friday meeting with White House chief of staff Susie Wiles
- The discussion centers on Mythos, Anthropic’s latest AI model featuring sophisticated cybersecurity functions
- Relations between Anthropic and the Pentagon became strained after disagreements over AI usage terms
- Multiple federal departments including Treasury and State have asked for Mythos briefings and system access
- Federal agencies are expected to receive a specialized version of Mythos, according to Bloomberg
Dario Amodei, CEO of Anthropic, will meet with Susie Wiles, White House chief of staff, at the West Wing this Friday. Sources with knowledge of the situation describe this as among Amodei’s most significant efforts to improve relations with the current administration.
The timing aligns with Anthropic’s preparation to launch Mythos, its newest AI system. Company officials have indicated the model presents cybersecurity challenges that could potentially trigger significant internet-wide disturbances.
On April 7, Anthropic unveiled Mythos, rolling it out via a restricted initiative named “Project Glasswing.” This program grants selected institutions preview access to the model exclusively for defensive cybersecurity applications.
White House officials have reportedly shown interest in Mythos’s sophisticated features. An individual familiar with ongoing discussions told Axios that denying the United States access to the technological benefits this model provides would be “grossly irresponsible” and would serve China’s interests.
The preview version has been distributed to technology firms and entities managing essential infrastructure. Public availability remains outside Anthropic’s current roadmap.
Federal Departments Request System Access
Both Treasury and State Departments have contacted Anthropic requesting presentations and entry to Mythos, the New York Times’ DealBook newsletter reports. Neither department’s representatives have issued immediate comments.
Thursday reporting from Bloomberg News indicates the federal government intends to distribute a Mythos variant across primary federal agencies.
Anthropic co-founder Jack Clark acknowledged Monday that company representatives have held talks about Mythos with administration officials, continuing even after the Pentagon terminated its business relationship with the AI laboratory.
Origins of the Pentagon Conflict
Tensions between Anthropic and the administration have persisted for several months. The core disagreement involves Pentagon requirements that Anthropic permit its Claude systems for “all lawful uses.”
Anthropic declined this demand, seeking specific safeguards preventing AI deployment in autonomous weaponry and mass surveillance operations. Defense Secretary Pete Hegseth subsequently designated the company as a security threat, prompting President Trump to order federal entities to sever connections with Anthropic.
The company has initiated legal challenges against government actions in two separate jurisdictions. Legal ambiguity persists for both parties.
Throughout this timeframe, reports suggest Anthropic’s systems have seen application during operations involving Iran. Neither White House officials nor Anthropic representatives have provided statements regarding Friday’s scheduled meeting.
