Dario Amodei, CEO of Anthropic, is reportedly set to meet Susie Wiles, Chief of Staff to President Trump, on Friday amid ongoing tensions following the Pentagon’s decision to blacklist the AI firm. This move came after Amodei criticized the Trump administration.
The meeting in the West Wing is anticipated as both U.S. and international leaders discuss Anthropic’s new artificial intelligence model, dubbed Mythos. The company claims it could significantly disrupt online security due to its advanced capabilities.
“It would be irresponsible for the government to disregard the technological advancements this new model could bring,” a source close to the situation mentioned, indicating that such a move could also benefit China.
Reports indicate that Anthropic is negotiating to offer government agencies early access to its systems. In a recent administration meeting, key figures, including Vice President J.D. Vance and Secretary of Defense Pete Hegseth, collaborated with Amodei to strategize around the Mythos model.
Instead of a broad public unveiling, Anthropic initiated “Project Glasswing.” This initiative aims to provide Mythos to a select group of about 40 companies, such as Amazon, Apple, CrowdStrike, Google, JPMorgan Chase, and Nvidia, enabling them to utilize the model for identifying and addressing security vulnerabilities.
Anthropic and the White House have experienced significant friction over safety regulations for months. Last month, the Pentagon labeled Anthropic a supply chain risk, prohibiting it from working with the Department of the Army and preventing contractors from applying its AI models in defense tasks. This was largely due to Anthropic’s refusal to remove safeguards meant to prevent its models from being used for autonomous weapons or mass surveillance within the U.S.
In a noteworthy incident, Amodei criticized Trump in a leaked email, alleging that the War Department was targeting Anthropic for not offering “dictator-style praise.” He later expressed regret as his company contested the blacklisting in court.
Despite these challenges, preparations are underway for major federal agencies to utilize versions of Mythos. Recently, staff from two significant agencies approached Anthropic about incorporating Mythos into their cybersecurity efforts.
Reportedly, Anthropic has engaged lobbyists with connections to the White House in an effort to ease tensions. Following its blacklisting, the company has brought on well-known lobbyist Brian Ballard, who has raised more than $50 million for Trump’s 2024 campaign, as well as Carlos Trujillo, a veteran who previously worked with Trump.
While the Pentagon has made agreements with competitors like OpenAI and xAI to deploy their models in classified contexts, experts believe it may take months before those systems reach the level of integration that Anthropic’s technology has already achieved.
