Congressman-elect Scott Wiener authored the country’s first major state AI safety law, and he’s used that cred to join a Democratic majority that appears ready to comprehensively regulate the fast developing technology for the first time on a national level.
Yet, he’s a freshman.
That presents a challenge. It’s unclear how the newcomer from San Francisco will truly flex his AI regulatory skills — rookie members do not usually get the most coveted and powerful committee assignments on Capitol Hill.
Still, his move to D.C. comes as Democrats resoundingly take control of the House with what some view as a mandate to craft comprehensive AI regulation that could reach President Donald Trump’s desk in the upcoming session.
The main question is whether he’d sign it.
“The president obviously may not care about what good we can accomplish here, but we’ll see what actually happens when push comes to shove,” said Rep. Zoe Lofgren, ranking member on the House Committee on Science, Space and Tech. “The midterms showed loud and clear that Americans want responsible action relating to AI.”
The House Committee on Energy and Commerce has been a main venue for AI safety fights, and rarely do first-term representatives get a spot. AI safety experts and lobbyists doubt Wiener will get on the panel, even as Democrats’ share of seats is expected to grow.
Instead, they say Wiener’s likelier route to serious regulation is through Lofgren’s House Committee on Science, Space and Tech, which is open to first-timers, or a new select committee House Democrats have been mulling specifically on AI.
“I welcome all excitement over joining our committee and I am excited to speak with prospective members about it,” Lofgren told Mission Local.
Wiener told MIT Technology Review that he would be eager to serve on the select committee on AI, if created. He declined to comment for this story.
Experts agree either panel would be a natural fit for Wiener.
“I would be jaw-dropped if Scott Wiener didn’t get the House Committee on Science Space and Tech,” said one AI safety lobbyist in D.C., granted anonymity to speak freely. “I think that’s the committee where he lands and provides him with the vehicle to get frontier safety bills on the table.”
The timing could not be more pertinent. The day after Wiener won the race to replace Nancy Pelosi, a team of outside AI evaluators released a jarring report claiming the San Francisco-based Anthropic’s own use of artificial intelligence had dramatically sped up the company’s research in the field.
More troubling: Because the review happened on Anthropic’s terms, Anthropic redacted major portions of the evidence reviewed by the Berkeley, California-based nonprofit research institute Model Evaluation and Threat Research, or METR.
Some publicly interpreted the report as evidence that Anthropic’s AI systems are improving themselves in a recursive fashion, a major risk for those who worry about AI models quickly refining themselves faster than humans can understand them. METR said AI is not yet able to make “highly significant and fully autonomous contributions” to the most important parts of the model training process.
Responses to the Nov. 5 report ranged from ho-hum to hysterical, underscoring the ways in which the risks of AI systems — like the loss of human oversight — remain difficult, even for experts, to parse.
“I need to call my family,” wrote Jake Mendel of the effective altruist philanthropy group Coefficient Giving, alarmed by the idea of “superintelligence” in the near future. “I feel sick,” wrote Dane Sherburn, who previously worked on OpenAI’s preparedness team and co-founded P-Zero Research. “I don’t know what evidence people will need to see before they start taking these risks seriously.”
There was also a fair amount of skepticism. “Looks to me more like a one-time speedup than continuous acceleration,” wrote George Mason University economist Robin Hanson.
Last year, Wiener authored successful state legislation laying out transparency and reporting requirements for AI models. While quick to move on regulation, he’s opposed more hardline stances, such as a moratorium on data centers.
His campaign saw millions of dollars in support from an Anthropic-backed AI safety nonprofit called Public First Action, and direct cash from Anthropic and OpenAI employees, many of whom have expressed concern over the technology while continuing to develop it.
Wiener has insisted that there was “no Anthropic money in this race,” as the company said it earmarked its giving to Public First Action for non-political purposes, but money is fungible.
AI safety advocates, meanwhile, are looking at Wiener for more than just a vote.
“Energy and commerce will most likely deal with the biggest AI issues — if I were him, that’s where I will look to go, but that’s a highly competitive one to get on,” said Sacha Haworth, executive director of the Tech Oversight Project.
Other questions remain as to how Democrats in power will handle AI regulation. Progressives may want to focus more on oversight investigations while moderates look to repair the party’s relationship with “big tech” and its money, said the unnamed D.C. lobbyist.
There’s also the federal question of preemption, which could void state laws — such as Wiener’s legislation — that have sprung up in the absence of a comprehensive federal law.
AI has affected so many aspects of public life, Haworth said it would not be difficult to find ways to legislate on AI from less obvious panels, too.
“The Veterans Affairs committee, for example, could do something on AI scams,” Haworth said. “AI is everywhere and there will not be a dearth of places to regulate them.”
