A federal agenda after the election
Wiener won the Nov. 3 midterm election in the district long represented by former House Speaker Nancy Pelosi, MIT Technology Review reported on Nov. 9. The California AI legislation author now enters federal policy debates as he warns that technological advances are bringing greater risks.
His proposals extend beyond a new regulator. According to The Hill, Wiener campaigned on drafting legislation covering AI "kill switches," safety standards, legal liability and international partnerships. He also pledged to expand unemployment benefits in anticipation of possible job displacement and use tax policy to share AI's benefits with the public.
Wiener said he would join a special congressional AI committee if one is established. But he expressed little optimism about the prospects for federal regulation, citing President Donald Trump's policy direction and Congress's legislative record, while pledging to keep pursuing effective safety measures.
He also opposes broad federal restrictions on states' power to regulate AI, arguing that states need legislative independence to respond quickly as the technology changes. On data center construction, he said public resistance reflects concerns about electricity costs, water consumption and neighborhood living conditions rather than AI's long-term risks.
The research informing his warning
Wiener cited an assessment by AI safety research group METR in warning that AI development could accelerate beyond manageable levels. The report put the pace of performance gains in Anthropic's models at about 1.9 times the previous rate and identified AI-assisted research and development as a significant contributor.
Published Nov. 5, 2026, the assessment drew on 11 weeks of high-access evaluation at Anthropic, from Aug. 12 through Oct. 29. METR estimated that Anthropic's own use of AI in R&D accounted for most of the current acceleration in its models' capabilities, perhaps about 80%.
METR attributed the acceleration to factors other than faster scaling of training compute and said it found no convincing bottlenecks likely to slow progress over the following 12 months. Separately, Anthropic employees reported productivity gains reaching about 12 times previous levels in October 2026.
Those findings do not establish that AI has achieved recursive self-improvement: the ability to develop a subsequent generation of models without human intervention.
California's law and its limits
Wiener's federal proposals follow two major legislative efforts in California. His advanced AI bill, SB 1047, failed after a gubernatorial veto in 2024. A revised measure with adjusted regulation, SB 53, became law the following year.
Gov. Gavin Newsom signed SB 53, the Transparency in Frontier Artificial Intelligence Act, on Sept. 29, 2025. It was expected to take effect Jan. 1, 2026. The law covers developers of frontier models trained using more than 10^26 integer or floating-point operations, with additional requirements for large frontier developers whose annual gross revenue, including affiliates, exceeds $500 million.
Large frontier developers must post a "frontier AI framework" on their websites. Frontier developers must notify the California Office of Emergency Services of critical safety incidents within 15 days of discovery, or within 24 hours when an incident presents an imminent risk of death or serious physical injury.
SB 53 also provides protections for whistleblowers reporting catastrophic risks and establishes CalCompute, a public cloud-computing cluster. California's attorney general can pursue civil penalties of up to $1 million for each violation.
The law became a point of contention in Wiener's House race. His opponent, San Francisco Supervisor Connie Chan, described it as watered down and said the AI industry "wanted weaker accountability." She accused Wiener of leaving most AI workers and contractors outside its whistleblower protections, cutting maximum penalties from $10 million to $1 million and removing independent audit requirements. Wiener disputed her account, saying the law "didn't get watered down."
Other states adopted similar transparency legislation after California: New York and Illinois both passed bills. Illinois state Rep. Daniel Didech said SB 53 substantially informed his measure, which Gov. J.B. Pritzker signed in July 2026.
Campaign support draws scrutiny
Financial support from an Anthropic-backed group has also brought scrutiny of Wiener's industry ties. Public First Action supplied substantial funding to a political action committee supporting him, prompting concerns about conflicts of interest.
Anthropic announced that it contributed $40 million to Public First Action in two installments during 2026. The group said on June 30 that it had raised $80 million.
The San Francisco Examiner reported that Public First Action gave $500,000 in April to Abundant Future, a PAC that spent against Wiener's primary opponent Saikat Chakrabarti. The PAC subsequently spent in support of Wiener and against Chan.
Wiener defended the support, saying it recognized his record of advancing AI regulation.
