Anthropic and the Pentagon: Claude Mythos Changes the Game

The Trump administration and the artificial intelligence company Anthropic have experienced a period of marked tension, with the government labeling Anthropic as a "radical left and woke company" and its members as "leftist lunatics." However, this tense situation could evolve with the launch of Anthropic's new cybersecurity model, Claude Mythos Preview.

Relations between Anthropic and the Pentagon deteriorated at the end of February. The company had refused to use its technology for controversial military applications, such as domestic mass surveillance or lethal autonomous weapons without human intervention. Despite these tensions, Anthropic was the first company to receive permission to operate its models on classified military networks. This refusal led to public insults and the classification of Anthropic as a "supply chain risk," resulting in a lawsuit to contest this designation and a temporary injunction suspending its ban.

In an effort to regain the trust of the U.S. government, Anthropic introduced Mythos Preview, a model announced with advanced capabilities for detecting security issues across nearly all major web browsers and operating systems. This model is touted as the company's most powerful to date and is currently available only for private access. It is designed to flag critical vulnerabilities in some of the most widely used internet infrastructures, allowing companies like Apple, Nvidia, and JPMorgan Chase, which have already adopted it, to address these flaws before they can be exploited by malicious actors.

The launch of Mythos Preview has prompted emergency meetings between U.S. banking leaders and Jerome Powell, the Federal Reserve Chair, highlighting the model's importance in the financial sector. The Trump administration also seems to be taking note of these developments. Anthropic has indicated that it has had "ongoing discussions with U.S. government officials regarding Claude Mythos Preview and its offensive and defensive cyber capabilities." Dianne Penn, the product management lead at Anthropic, confirmed that the company had briefed senior U.S. government officials on Mythos and its capabilities, while remaining committed to working closely with all levels of government.

Meanwhile, Anthropic recently hired Ballard Partners, a lobbying firm linked to Trump, fueling speculation about a potential deal with the White House. On Friday, Axios reported that Dario Amodei, CEO of Anthropic, was scheduled for a meeting with White House Chief of Staff Susie Wiles. A source close to the negotiations stated that it would be "grossly irresponsible for the U.S. government to deprive itself of the technological advancements presented by the new model" and that "it would be a gift to China."

Some segments of the U.S. intelligence community, as well as the Cybersecurity and Infrastructure Security Agency (CISA), are already testing Mythos Preview, and other departments and agencies are showing increasing interest. If Amodei did indeed attend the meeting, it could pave the way for further integration of Claude into government use across agencies, potentially influencing the DoD's stance on the use of Anthropic's technologies.