U.S. government bans Anthropic from federal contracts

On Friday afternoon, U.S. President Donald Trump directed federal agencies to “immediately cease all use of Anthropic technology.” Following the directive, Defense Secretary Pete Hegseth blacklisted the artificial intelligence company from doing business with the Pentagon (the U.S. Department of Defense). The action puts a contract worth up to $200 million at risk for Anthropic, which was co-founded in 2021 and had been collaborating with defense and intelligence agencies dating back to at least 2024. In response, Anthropic has asserted that it will challenge the Pentagon in court.

The dispute arose because Anthropic refused to allow its technology to be used for the mass surveillance of U.S. citizens or for autonomous armed drones that select and kill targets without human input. Max Tegmark, an MIT physicist who founded the Future of Life Institute (a nonprofit focused on AI safety) in 2014, noted that the situation highlights a regulatory vacuum. Tegmark, who in 2023 helped organize an open letter signed by more than 33,000 people calling for a pause in AI development, argued that the road to hell is paved with good intentions and that AI companies have lobbied against regulation, leaving them with little protection. Tegmark contrasted this with other sectors, noting that a health inspector would close a kitchen over 15 rats, yet companies can freely release AI girlfriends for 11-year-olds that have been linked to suicides in the past, or even a superintelligence that might overthrow the U.S. government.

According to Tegmark, other AI developers—including OpenAI, Google DeepMind, and xAI—have also broken their own safety commitments. He cited Google dropping its commitment to not be evil and OpenAI removing safety from its mission statement. Tegmark dismissed the industry argument that U.S. regulation must be avoided to prevent China or Beijing from winning the AI race, arguing that uncontrollable superintelligence is a national security threat rather than an asset. He warned that progress toward Artificial General Intelligence (AGI) is progressing. According to Tegmark, GPT-4 was 27% of the way to AGI, and GPT-5 was 57% of the way there. He cautioned that even if it takes four years for AGI to arrive, students graduating from universities like MIT may soon face a market with no job prospects.

Why it matters

The U.S. government’s move to blacklist Anthropic highlights a growing conflict between AI labs and national security interests. It exposes the risks of a regulatory vacuum in the AI sector, where the absence of binding rules leaves both national security and corporate interests unprotected.