Anthropic CEO Dario Amodei met with top Trump administration officials on Friday to discuss the responsible deployment of the company’s new AI model, Mythos. As detailed by the Wall Street Journal, the meeting marks a significant step in easing tensions between Anthropic and the federal government.
Among those present were White House chief of staff Susie Wiles and Treasury Secretary Scott Bessent, who has warned financial industry executives about the potential cyber threats associated with Mythos. The White House described the discussion as “both productive and constructive,” with talks focusing on “opportunities for collaboration, as well as shared approaches and protocols to address the challenges associated with scaling this technology.”
Anthropic has already provided a preview version of Mythos to major tech companies and organizations responsible for critical infrastructure, and has been briefing government officials on strategies to minimize potential harms. The company currently has no plans to release the model to the general public and is in talks to grant government agencies advance access.
The feud between Anthropic and the White House has been months in the making
The recent collaboration represents a thaw in a bitter, months-long dispute rooted in the Pentagon’s demand that Anthropic’s Claude models be available for “all lawful uses.” Anthropic refused, insisting on explicit protections against the use of its AI for autonomous weapons and mass surveillance. That standoff led Defense Secretary Pete Hegseth to label Anthropic a supply-chain risk, a designation that had never previously been applied to an American company, and President Trump subsequently directed federal agencies to cut ties with Amodei’s firm.
The DoD does maintain its own guidelines on autonomous systems, codified in Directive 3000.09, which was updated on January 25, 2023. The directive requires that autonomous and semi-autonomous weapon systems be designed to preserve appropriate human judgment over the use of force, that operators adhere to the law of war and established rules of engagement, and that all AI-enabled systems align with the DoD’s AI Ethical Principles and Responsible AI Strategy. Anthropic nonetheless sought additional, more specific safeguards beyond those existing standards.
The dispute escalated into legal battles across two courts, even as Anthropic’s technology continued to be deployed and used during the conflict in Iran. Security analysts raised concerns that penalizing one of the country’s leading AI firms was counterproductive, particularly as it prepared to release models with significant cybersecurity implications.
Mythos appears to have provided the opening for de-escalation. Earlier this week, a top official at the Office of Management and Budget notified government agencies of protections being put in place to allow them to begin using the model, an effort coordinated by National Cyber Director Sean Cairncross. The administration is also working with other leading AI developers to help secure critical software vulnerabilities.
On the lobbying front, Anthropic hired Brian Ballard, known for raising over $50 million for Trump in the 2024 election, after the supply-chain risk designation. The company also works with a firm founded by Carlos Trujillo, who served in the first Trump administration. Amid those efforts, analysts have cautioned that Anthropic’s combative approach with the Defense Department could complicate its ambitions to become a top government model provider, particularly as it considers going public.
The Pentagon has already forged deals with OpenAI and Elon Musk’s xAI to use their models in classified settings, with senators raising concerns about xAI’s Pentagon access and the safeguards around Grok’s deployment. Security analysts believe it will likely take months before those alternative models are as deeply embedded in operations as Anthropic’s technology has been.
The European Union is also in talks with Anthropic about Mythos, though it does not currently have access to the model. A European government official noted that Mythos became an unplanned agenda item for international leaders convening in Washington during the International Monetary Fund’s annual meeting this week, with leaders looking to develop a multilateral response to the cyber risks posed by the model.
Published: Apr 17, 2026 07:45 pm