TLDR:
- OpenAI supports California bill AB 3211 requiring ‘watermarking’ of AI-generated content
- Elon Musk backs California bill SB 1047 to regulate AI, breaking with some tech giants
- SB 1047 would require large-scale AI models to undergo safety testing
- The bill has faced criticism from some tech companies and lawmakers
- California lawmakers have introduced multiple AI-related bills this legislative season
California’s efforts to regulate artificial intelligence have ignited a heated debate in Silicon Valley, with tech giants and industry leaders taking opposing sides on two key bills.
The state’s lawmakers have introduced dozens of AI-related measures this legislative season, but two bills, in particular, have garnered significant attention: AB 3211 and SB 1047.
AB 3211, supported by ChatGPT developer OpenAI, would require tech companies to label AI-generated content.
This bill aims to increase transparency and help people understand the origin of content they encounter online, especially in light of the numerous elections taking place globally this year.
OpenAI’s Chief Strategy Officer, Jason Kwon, emphasized the importance of such measures in a letter to California State Assembly member Buffy Wicks, who authored the bill.
On the other hand, SB 1047, backed by Elon Musk, has sparked more controversy. This bill would mandate safety testing for large-scale AI models and implement various safeguards.
Companies developing these models would be required to agree to third-party audits and implement a “kill switch” to shut down the technology if necessary. The bill also proposes protections for whistleblowers.
Musk’s support for SB 1047 puts him at odds with some of Silicon Valley’s most influential players, including Marc Andreessen and OpenAI. Critics argue that the bill’s requirements are vague and burdensome, potentially stifling innovation and open-source development.
However, supporters, including the bill’s author, Democratic state Senator Scott Wiener, contend that such measures are necessary to prevent potential catastrophic harm to humanity.
The tech industry’s response to these bills has been mixed. While OpenAI supports AB 3211, it has expressed reservations about SB 1047, arguing that AI regulations should be decided at the federal level. Google, Meta, and Andreessen Horowitz have also voiced concerns about SB 1047, viewing it as a threat to innovation and research.
Some prominent figures in the AI field have weighed in on the debate. Dr. Fei-Fei Li, co-director of the Stanford Institute for Human-Centered Artificial Intelligence, warned that SB 1047’s penalties and restrictions could have unintended consequences on innovation, particularly for the public sector, academia, and smaller tech companies.
The bills have also drawn attention from federal lawmakers. Eight California House members, including former House Speaker Nancy Pelosi, have urged Governor Gavin Newsom to veto SB 1047 if it passes. They argue that while well-intentioned, the bill is ill-informed and could potentially harm California’s AI ecosystem.
Despite the opposition, Senator Wiener defends SB 1047 as a “reasonable” and “light touch” approach to AI regulation. He argues that many of the bill’s requirements align with commitments already made by major tech companies. Wiener also emphasizes that the bill has been amended to address some concerns, including the removal of criminal perjury charges for lying about AI models.
With the legislative session coming to a close, both bills face crucial votes in the coming days. If passed, they would advance to Governor Newsom for final approval. The governor has not yet indicated his stance on either bill, leaving the fate of California’s AI regulation efforts uncertain.
