Emil Michael, who previously led the Pentagon's negotiations with artificial intelligence company Anthropic regarding military AI safety guardrails, has been named one of TIME's 100 Most Influential People in AI. Michael's tenure at the Pentagon, which began last August, has been marked by efforts to streamline the adoption of AI technologies for military use.
Michael, a former executive at Uber, aimed to revise what he described as "25 pages of restrictions" inherited from previous Biden-era contracts that hindered basic military activities. He advocated for terms that would permit "all lawful uses" of AI models by the Pentagon.
The dispute with Anthropic centered on two specific exceptions: the company's AI, Claude, not being used for mass surveillance of Americans and not independently controlling lethal autonomous weapons. Michael has disputed Anthropic's account, stating that the company sought to have the Pentagon require its approval for specific uses of Claude, which he characterized as an attempt to insert Anthropic into the military chain of command. Anthropic has denied this assertion, calling it "patently false."
Following the breakdown of negotiations, the Pentagon designated Anthropic as a supply chain risk, a label previously not applied to an American company. Michael commented on the designation, suggesting that Anthropic's public engagement on the issue contributed to it. "When you do that, you are a supply chain risk," he told TIME.
Despite the unresolved disagreement with Anthropic, Michael's broader objectives appear to have advanced. Following the Pentagon's actions, other major AI developers, including OpenAI, xAI, and Google, reached agreements with the Department of Defense that allowed for the deployment of their models under terms more aligned with Michael's outlined conditions.
Investors in Anthropic include Salesforce, where TIME owner and co-chair Marc Benioff serves as CEO.