Report from the New York Times
In Brief – California has enacted legislation imposing major new transparency requirements on AI developers. The “Transparency in Frontier Artificial Intelligence Act” (TFAIA) requires developers of the largest AI models to meet the most stringent standards and creates new whistleblower protections for employees at AI labs. California Governor Gavin Newsom signed the law in the face of opposition from the state’s AI, technology and venture capital industries, who argued that premature regulation of a rapidly evolving technology could hamper its development. Anthropic was the only major AI developer that publicly supported the bill. OpenAI’s recommendation to the Governor was for the state to consider developers to be compliant with the state’s safety rules if they meet federal or European standards, especially with an eye on implementation of the EU AI Act that is plodding forward in Europe.
Context – AI regulation notched landmark wins in 2024, especially with the EU’s AI Act. Colorado also became the first US state to enact broad AI regulation, and the UK and Japan appeared to be moving in that direction. However, momentum shifted last fall. President Trump staked out a strong pro-industry campaign position which is now reflected in his anti-regulatory AI Action Plan. California passed a slew of AI-related bills in 2024. The highest profile was SB 1047, “AI safety” legislation often compared to the EU AI Act. Governor Newsome vetoed it. He did sign two so-called political “deepfake” bills that have been blocked by a federal judge. The UK and Japan have since stepped back from AI regulation, and the pace of the EU’s AI rules is now being criticized by European business leaders and noteworthy public officials. This summer, the big Republican budget reconciliation bill initially included a 10-year moratorium on restrictive state AI laws, but it proved to be a deregulatory bridge too far and was pulled from the bill. However, backers continue to argue that state regulation, especially by California, will slow AI advancement in the country. New York’s governor is still pondering whether to sign state-level AI regulation into law.