one year on
Anthropic Endorses California AI Transparency Bill SB 53
The frontier lab breaks with much of the tech industry to back state-level transparency rules, arguing powerful AI cannot wait for federal action.
Originally published Published on this site
Anthropic today publicly endorsed California’s SB 53, a bill by state Senator Scott Wiener that would impose the nation’s first transparency requirements on frontier AI developers. In a blog post, the company said it still prefers federal rules but argued that “powerful AI advancements won’t wait for consensus in Washington.” The endorsement is a rare and significant win for the bill, which requires large AI developers to publish safety frameworks, release transparency reports before deploying new models, report critical safety incidents within 15 days, and protect whistleblowers. Companies that fail to comply could face financial penalties.
The bill targets the largest AI model developers, particularly those with gross revenue of more than $500 million, and uses a compute threshold of 10^26 FLOPS to determine which systems are covered. It was shaped by an expert panel convened by Governor Gavin Newsom after his veto of last year’s broader SB 1047. Anthropic says SB 53 largely codifies practices it already follows—like its Responsible Scaling Policy and system cards—and would create a level playing field where disclosure is mandatory, not optional. The company acknowledged the bill has room for improvement, particularly in updating compute thresholds and requiring more detailed evaluations.
Much of the tech industry opposes SB 53. The Consumer Technology Association and Chamber of Progress are lobbying against it, and Andreessen Horowitz has argued it may violate the Commerce Clause. OpenAI’s chief global affairs officer, Chris Lehane, sent the governor a letter in August warning against regulations that could push startups out of California. But policy experts say the bill’s more modest approach—dropping earlier third-party audit requirements—gives it a realistic chance of passage. Dean Ball, a former White House AI policy adviser who criticized SB 1047, said the drafters have “shown respect for technical reality” and “legislative restraint.” The bill still needs a final Senate vote before reaching Newsom’s desk.
The record
Anthropic co-founder argues on X that the industry can't wait for federal government and that SB 53 creates a solid blueprint for AI governance.
OpenAI's chief global affairs officer sent a letter to Governor Newsom in August arguing against AI regulation that could push startups out of California, though not mentioning SB 53 by name.
Former OpenAI head of policy research said on X that Lehane's letter was 'filled with misleading garbage about SB 53 and AI policy generally.'
Andreessen Horowitz's AI policy head and chief legal officer published a blog post arguing that state AI bills risk violating the Constitution's Commerce Clause.
One year later — open only if you can handle spoilers
Newsom signed SB 53 into law on September 29, 2025 as the Transparency in Frontier Artificial Intelligence Act, making California the first US state to require frontier AI developers to publish safety frameworks, file transparency reports, and report critical safety incidents; its core provisions take effect January 1, 2026. Anthropic's endorsement — a rare break from an industry that had lobbied hard against the bill, with OpenAI, Meta, and venture firms like Andreessen Horowitz opposing or seeking to weaken it — was widely credited with helping SB 53 clear the legislature a year after Newsom vetoed the broader SB 1047. By mid-2026 the fight had shifted to whether other states and Congress would adopt or preempt California's template.
Filed
Compiled and edited by Adam Allcock.
The Weekly Replay · free by email
This week, one year ago — every Sunday.
One email each Sunday: the week's replayed AI news, with the one-year-later annotations included. Written like it's breaking — dated like it isn't.
Free · double opt-in · unsubscribe anytime · privacy