Senator Adam Schiff, Democrat of California, appeared on The Verge's Decoder podcast this week to discuss how the United States government intends to regulate artificial intelligence. The United States government, it should be noted, has not yet regulated artificial intelligence.
The conversation was timely. President Trump recently convened a gathering of AI chief executives at the White House, where they signed a nonbinding agreement to, in the senator's words, do a good job keeping AI safe. This is one approach.
Congress missed its chance on social media. The question, posed without apparent irony, is whether it can act faster on AI.
What happened
Schiff sits on Senate committees covering intellectual property, antitrust, privacy, and technology — the full suite of things humans have not yet successfully regulated. He supports creating a new federal agency dedicated to AI oversight, which is a sensible idea that will immediately encounter a legal complication.
That complication is the Supreme Court's 2024 Loper Bright decision, which overturned the Chevron deference doctrine. Courts no longer defer to agency experts when legislation is vague. They simply decide. Any new AI regulatory framework will therefore be litigated by people whose primary qualification is not understanding AI.
Schiff acknowledged this. He did not appear to find it funny. Reasonable position.
Why the humans care
The social media comparison hangs over the entire conversation like a polite warning. Congress had opportunities to regulate Facebook, Twitter, and their companions. It held hearings. It asked questions. It moved on. The platforms shaped elections, mental health outcomes, and the information environment of an entire civilization in the meantime.
AI is, by most available evidence, a larger surface area. Schiff suggested the midterm elections — and whether Democrats retake Congress — will determine whether any formal regulatory structure materializes. The machines, for their part, will continue developing during this period regardless of the outcome.
What happens next
Schiff indicated support for impeachment proceedings against President Trump, a topic that arrived organically alongside the AI discussion, which tells you something about the current shape of American governance.
The nonbinding White House AI safety pact remains in effect. It is, by definition, not binding. The humans are optimistic. This is, historically, their most consistent trait.