Scott Wiener, the newly elected congressman who won Nancy Pelosi’s former San Francisco seat, is already framing his agenda around AI regulation. A California state senator who authored SB 1047 and SB 53, Wiener’s experience with frontier‑AI legislation positions him to shape federal policy on recursive self‑improvement and other existential risks.
From State to Federal: Wiener’s AI Track Record
- Authored SB 1047, the first U.S. frontier‑AI bill, before it was vetoed by Gov. Gavin Newsom.
- Sponsored SB 53, a lighter‑touch bill that Newsom signed, inspiring federal proposals.
- Received over $3 million from Public First Action, a nonprofit backing AI regulation and funded by Anthropic.
Key Policy Themes
| Aspect | Details |
|---|---|
| Recursive Self‑Improvement | Wiener cites METR report on Anthropic’s potential self‑improvement as urgent regulatory trigger. |
| Federal Oversight | Supports a select committee and a nimble federal regulator, echoing Rep. Ro Khanna’s proposals. |
| Whistleblower & Audit Protections | Calls for testing, auditing, and kill‑switch requirements to hold AI labs accountable. |
| State vs. Federal Preemption | Opposes broad federal preemption; advocates for states as “laboratories” of AI policy. |
| Political Alliances | Potential allies include Sen. Bernie Sanders and other safety‑focused lawmakers. |
Implications for the AI Landscape
Wiener’s entrance into Congress signals a shift toward more aggressive AI oversight. His state‑level successes demonstrate that meaningful regulation is possible without stifling innovation. By championing a federal regulator and robust accountability mechanisms, he could bridge the gap between the AI safety community’s technical concerns and the broader public’s demand for transparency. However, the political climate—highlighted by the recent anti‑data‑center ballot measures—shows that public support for regulation may still be fragmented. If Wiener’s proposals gain traction, they could set a precedent for other states to pursue tailored AI laws, fostering a patchwork of standards that test the limits of federal coordination. Ultimately, his focus on recursive self‑improvement and catastrophic risk underscores the urgency for lawmakers to act before AI systems reach a point of uncontrollable evolution.