Anthropic chief 'productive' White House talks despite Trump blacklist after refusing to use AI for military purposes
The Trump administration and Anthropic are seeking a compromise after the AI company was blocked for refusing to cooperate with the Pentagon. At stake in the talks is access to the new cybersecurity tool “Mythos,” according to CNN.
Dario Amodei, chief executive of Anthropic, met with Chief of Staff Susie Wiles on Friday as Anthropic is in litigation with the US government. The White House described the discussion as “productive,” aiming for “common protocols for the challenges of this technology.”
“We discussed opportunities for collaboration, as well as common approaches and protocols to address the challenges of expanding this technology,” the White House said in a statement issued Friday afternoon.
“The discussion also addressed the balance between promoting innovation and ensuring safety. We look forward to continuing this dialogue and will hold similar discussions with other leading AI companies.”
Anthropic confirmed the nature of the discussions, stating that the main topic was “cyber security, America's leadership in the AI race and AI safety.”
“The meeting reflected Anthropic's continued commitment to working with the US government on the development of responsible AI,” a company spokesperson said.
Until recently, Anthropic's Claude model was the only AI model available on the Pentagon's classified network. But Trump recently announced that his administration would cut ties with the company after Anthropic refused to make concessions on terms that would have allowed the military to use the Claude model for “all legal purposes,” including autonomous weapons and mass surveillance.
After negotiations over the use of the Claude system failed, the Pentagon declared Anthropic a “supply chain risk,” a label previously used only for companies associated with foreign adversaries. This measure would have practically led to the exclusion of the Anthropic company from the list of government suppliers.
In response, Anthropic sued the administration. Although a California judge initially blocked the measure, the D.C. Circuit Court of Appeals later ruled that the Defense Department can cut ties with the supplier during the trial. The court cited the need not to force the military to work with an unwanted supplier “in the midst of a large-scale military conflict.”
The company claims that its models are not yet reliable enough for autonomous weapons and that the legislation does not protect citizens in the event of mass surveillance. In this context, Anthropic hired the lobbying firm Ballard to manage the relationship with the War Department.
Meanwhile, Anthropic announced an upcoming powerful AI model called Mythos, which both the company and experts warned could be a “watershed” moment for cybersecurity, allowing certain groups to gain early access to assess their cybersecurity risk.
The Office of Management and Budget has already told federal agencies that it is preparing to give them access to Mythos for testing. The White House is also in talks to gain access to the model, but Anthropic declined to comment on whether the Trump administration is already working on testing the system.
“Any new technology that may be used or implemented by the federal government requires a technical evaluation period to verify its accuracy and security,” a White House official said. He added that the collective effort will help fix critical software vulnerabilities.
