Pentagon official laments some AI companies’ reluctance to fully commit to military’s operations

WEST PALM BEACH, Fla. — It is “disheartening” that some cutting-edge tech companies seem reluctant to fully do business with the military and support all of its operations, a key Defense Department official said Tuesday amid an escalating feud between the Pentagon and Silicon Valley firm Anthropic over the reported use of the company’s AI tool in recent U.S. Special Forces missions in Venezuela.

Emil Michael, the Pentagon’s under secretary of war for research and engineering, told an audience in Florida that the nation is at a pivotal moment that requires a new level of cooperation between the military and leading companies, including those in the tech sector.

Mr. Michael spoke at the inaugural Defense Tech Leadership Summit, sponsored by Amazon Web Services, the real estate development company Related Ross, and Vanderbilt University.

Mr. Michael drew a parallel to 2018, when thousands of Google employees vehemently and publicly objected on ethical grounds to their company’s involvement in the Pentagon’s controversial Project Maven, which sought to leverage AI to better identify targets for drone strikes. Mr. Michael said that he believes the company later regretted the outspoken opposition many of its employees expressed to the program.

“It’s a little disheartening that in this day and age there are still a few companies that aren’t thinking in that mindset and don’t remember those lessons of the past, because we’re in a much different threat environment now relative to where we were in 2018. We’re more in a 1938 moment where the defense industrial base has to be rebuilt,” Mr. Michael said, referring to the final months before the outset of World War II in 1939.

The nation’s defense industrial base and the companies that compose it, he said, are a crucial part of the broader American national security ecosystem.

“A lot of what we’ve done in the last 25 years has been somewhat neglectful of importing tech that’s been done in the private sector into the department to make it more efficient, more lethal, more effective,” Mr. Michael said. “You want the Department of War to create a sense of deterrence, that [an enemy] can’t win, so don’t try. And that requires some technological … importation of the best capabilities.”

The comments from Mr. Michael — a Silicon Valley veteran — come amid reports that Defense Secretary Pete Hegseth is considering cutting all Pentagon ties with Anthropic, creator of the popular ClaudeAI tool. U.S. forces reportedly used Claude in the operation to capture Venezuelan President Nicolas Maduro in Caracas last month. 

Anthropic says its usage guidelines prohibit Claude from being used to facilitate violence, develop weapons or conduct surveillance. 

The unfolding dispute at the highest levels of the federal government and the nation’s increasingly powerful tech sector underscores the thorny, complex questions at play when the U.S. military uses a company’s product in a lethal military mission. 

Axios first reported that U.S. forces used Claude in the preparation for the Maduro mission and while it was being conducted. The outlet reported Tuesday that Mr. Hegseth is considering cutting all Pentagon business ties with Anthropic and potentially labeling the company a “supply chain risk,” which would mean that any other firm seeking to do business with the Defense Department would have to sever its own partnerships with Anthropic. 

The Pentagon currently has contracts with Anthropic worth a reported $200 million.

An Anthropic spokesperson told Axios the two sides are trying to resolve the dispute.

“We are having productive conversations, in good faith, with DoW on how to continue that work and get these new and complex issues right,” the spokesperson said.

In his address Tuesday, Mr. Michael did not directly mention Anthropic. But his comments left little doubt about the growing frustration inside the Pentagon about what some officials view as unreasonable demands being made by private companies, particularly those in the AI space, about how their products are used.

“You may not want to help the Department of Defense, but you’ve got to see what the global threat is,” Mr. Michael said, drawing a contrast between the U.S. and its chief adversary, China, over how the two nations view AI in a military context and what kinds of moral outcomes that could lead to.

“They don’t trust their chain of command like we do. And therefore what fills that chain of command when you make an order? AI,” he said. “And their risk tolerance is a lot higher than ours.”

Under Mr. Hegseth, Mr. Michael and other Trump administration officials, the Pentagon is undertaking a major reimagining of the U.S. defense industrial base and the military’s work with private companies. Across the Defense Department and military services is an approach increasingly tilted toward commercial capabilities and how they can be incorporated into the armed forces.

That differs from past approaches in which the Pentagon often employed a defense firm to essentially build an entirely new capability specifically for military use.

Current and former high-ranking military leaders say that having the world’s best defense industrial base is itself a deterrent to adversaries.

“If you want to think about deterrence … the defense industrial base is a deterrent element,” said retired Adm. Christopher Grady, former vice chairman of the Joint Chiefs of Staff, at Tuesday’s Defense Tech Leadership Summit.

That blurring of the lines between commercial and military capabilities, and how companies are responding to it, will be a critical issue for the military and its private sector partners in the years to come.

“There are really great companies out there who have a ton of intellectual capital, who are developing amazing products, and I think they are starting to realize just how critical it is that this isn’t just a commercial situation anymore, right?” said Dave Levy, vice president of worldwide public sector for Amazon Web Services, at Tuesday’s forum.