Anthropic Sues Trump Administration Over AI Controversy
On Monday, Anthropic, a tech company based in San Francisco, filed a lawsuit against the Trump administration. The suit alleges that the administration effectively blacklisted AI companies that aimed to stop the Pentagon from utilizing chatbots for mass surveillance and weapons development.
Anthropic specifically held Army Secretary Pete Hegseth accountable for labeling the company as a supply chain risk—this designation is significant, as it marks Anthropic as the first U.S. company to receive such a label. The company argues this was retaliation for the Pentagon’s attempts to limit the use of Claude chatbots.
In their filed complaint, Anthropic described the action as “unprecedented and illegal,” stating, “The Constitution does not allow the government to use vast powers to punish companies for protected speech.”
The Pentagon, however, chose not to comment, citing that they do not address ongoing legal matters.
This legal action follows a recent apology from Anthropic’s CEO, Dario Amodei, regarding a leaked 1,600-word document that criticized the Trump administration harshly. Nonetheless, the company felt compelled to challenge the supply chain risk designation in court.
Amodei expressed regret over the “tone” of that fiery letter, in which he accused the Army Department of singling out Anthropic for not showing enough support for Trump.
In a memo sent last Thursday, he noted, “I would also like to directly apologize for the internal posts that were leaked to the media yesterday.” He clarified that Anthropic did not leak the document nor did they instruct anyone to do so, emphasizing, “It is not in our interest to escalate this situation.”
The comments from Amodei came shortly after Trump labeled Anthropic’s employees as part of “the left’s crazy business,” and Hegseth announced plans to mark the company as a supply chain risk.
Amodei reflected, “It was a difficult day for the company. I apologize for the poor quality of my posts. It does not reflect my careful or considered views.”
Analyst Dan Ives from Wedbush Securities remarked that the Pentagon’s supply chain risk label, a designation previously reserved for foreign companies viewed as national security threats like Huawei, is essentially a “scarlet letter” for Anthropic.
This label means defense contractors would have to confirm they aren’t using Anthropic’s AI models in their government contracts. It’s still uncertain whether this will lead to broader restrictions, especially since Hegseth previously indicated that Anthropic would be banned from any commercial activities with federal government partners, including major companies like Lockheed Martin, Amazon, and Google.
However, according to Amodei’s recent memo, Anthropic asserts that the “vast majority” of their clients will not feel the impact of this designation.
The company secured a $200 million contract with the Department of Defense back in July, becoming the sole supplier of AI models for government classified networks.
Despite this, Hegseth criticized Anthropic for seeking exemptions during contract discussions about mass surveillance and weaponry, arguing that the Pentagon should be free to use AI for any lawful purpose.
OpenAI subsequently received a contract to provide AI services to the Pentagon.
In the same memo, Amodei suggested that Anthropic was facing backlash because they didn’t “donate to Trump,” contrasting with “OpenAI/Greg” who made a significant contribution, referring to OpenAI’s President, Greg Brockman.
Interestingly, Amodei had donated to Kamala Harris’ unsuccessful presidential campaign and accused both OpenAI and the Department of Defense of attempting to tarnish Anthropic’s reputation.
He stated that the Department was spreading misinformation, claiming, for instance, that the contract specifics from OpenAI were never shared with Anthropic.
Amodei characterized OpenAI’s CEO, Sam Altman, as presenting himself as someone intent on seeking fair contract terms for everyone while secretly collaborating with the Army Corps to replace Anthropic once it was classified as a supply chain risk.
He concluded that OpenAI’s safeguards are “probably 20% real and 80% safe.”
During Morgan Stanley’s technology conference last Thursday, Altman responded to the criticisms, asserting that government authority should supersede that of private companies. He warned it could negatively impact society if businesses began to turn away from the democratic process, simply because some officials are not favored at the moment.
Yet, Altman also admitted that the timing of OpenAI’s deal, which came just hours after negotiations with Anthropic fell through, “appeared opportunistic and sloppy.”
