A senior U.S. Defense Department official has pushed back against intensifying concerns about artificial intelligence risks, following a public warning from a former engineer at Anthropic PBC that AI industry insiders believe the technology "could kill us all by 2030." Emil Michael, the Pentagon's Chief Technology Officer and top official for research and engineering, addressed the issue during a speech on Thursday, referencing a social media post by former Anthropic engineer Jacob Coxon that has garnered over 150 million views, alongside worries about a "dystopian" future dominated by unstoppable AI systems.

"It's easy to get caught in this doom loop of 'we're going to put 40% of Americans out of work, we're going to destroy humanity,'" Michael said at a defense industry conference in Washington. "This is really just a fear of data centers, a fear of change — all these anxieties converging into one well-written tweet." Michael, a former Uber executive who now leads the Pentagon's push to accelerate AI adoption within the U.S. military, argued that potential AI harms highlighted in posts like Coxon's can be mitigated through free markets, broad industry collaboration, and engagement with the government.

The Defense Department's AI push comes amid a bitter dispute with Anthropic, which had demanded additional safeguards on military use of its AI tools. Michael was instrumental in leading negotiations with the company, but talks collapsed earlier this year after the Pentagon rejected Anthropic's conditions, calling them an infringement on military authority, and subsequently designated the company as a threat to the U.S. supply chain. The conflict has since moved to federal court, with Anthropic filing lawsuits in California and Washington, D.C., seeking to overturn the Pentagon's designation. A judge in the California case has ruled in the company's favor, but the federal appeals court in Washington has yet to issue a decision.

Michael stated that the case before the Washington Circuit Court of Appeals is "more significant" and maintained that, without a preliminary injunction, Anthropic still poses a supply chain risk. Anthropic did not immediately respond to a request for comment on Michael's remarks from Thursday. Speaking to reporters afterward, Michael added that the Defense Department has already shifted approximately 90% of its classified AI workloads away from Anthropic, with plans to complete the remaining transition by the end of the month.