Senator Adam Schiff, a California Democrat, said a voluntary artificial intelligence safety pledge signed by technology executives at the White House this week is hard to reconcile with the national security risks he sees in advanced AI systems. Schiff made the comments in an interview with The Verge's Decoder podcast published Oct. 5, 2026, describing a gap between what industry leaders say publicly and what they agreed to inside the White House.
Schiff said the pledge is non-binding and that the executives who signed it have separately called for regulation, urging that the public not simply trust companies to act responsibly. He said the companies appear unwilling to draw the anger of a president he described as vengeful and will say what President Trump wants while in his presence. He pointed to the administration's gathering of AI chief executives as the backdrop for the agreement.
The senator, who sits on Senate Judiciary subcommittees covering intellectual property, privacy and technology, and antitrust, said AI models that advance through recursive processes leave observers with less understanding of how they make decisions or what they try to access lawfully. He cited the Pentagon's use of AI as a security concern and called for investigation into what role AI may have played in target selection in the bombing of an Iranian girls' school. He also said the administration has moved against Anthropic because the company wanted to avoid using AI for domestic mass surveillance or fully autonomous weapons systems.
On intellectual property, Schiff said he and Senator Curtis are working together to first establish what copyrighted and protected material was used to train AI models. He said that would allow a discussion about what qualifies as fair use, and that companies must retain that information. He noted the industry is not uniform, with streaming and technology companies differing in how much their film libraries matter relative to their AI portfolios.
Schiff said he supports creating a new agency to regulate AI and discussed how the Supreme Court's decision overturning Chevron deference, in a case known as Loper Bright, affects that idea. The ruling ended a Reagan-era precedent under which courts deferred to agency experts when laws were vague. Schiff said courts now decide such questions, which could mean any new regulation is immediately tied up in litigation.
The interview also covered whether Schiff would support impeaching Trump again over what he described as corruption, as well as how AI regulation might compare with the social media regulation Congress ultimately produced. Schiff said the administration justifies many actions on national security grounds that cannot be questioned, including its claim that a triumphal arch serves a security purpose because ammunition would be stored in it. He said the administration and Republican colleagues in Congress have not focused on the cost of living or improving people's lives.
More AI news from TechManNews.
