OpenAI, Microsoft, and Adobe Back California’s AI Watermarking Bill

OpenAI, Microsoft, and Adobe are publicly supporting the AB-3211 bill that would require tech companies to label AI-generated content in California.

Watermarks must be added to the metadata of photos, videos, and audio recordings that were produced or significantly modified by generative AI systems, according to the legislation.

The bill will also require large social media platforms, such as X and Instagram, to apply “easy-to-understand” labels to uploaded content that users can click to view. Visual labels should be applied to audio clips and music videos.

OpenAI penned its support in a letter to California State Assembly member Buffy Wicks, which was seen by Reuters. Adobe and Microsoft also backed the bill in similar letters viewed by TechCrunch.

“New technology and standards can help people understand the origin of content they find online, and avoid confusion between human-generated and photorealistic AI-generated content,” OpenAI Chief Strategy Officer Jason Kwon wrote in his letter, according to Reuters.

Kwon’s letter also highlighted that transparency and requirements around provenance are especially important in an election year, as convincing synthetic content could be generated to swing votes. For example, AI-generated images of Taylor Swift fans endorsing Donald Trump were circulated on social media in early August.

AB-3211 has been approved by the state Assembly and senate appropriations committee. A full state Senate vote will come next before it is advanced to Governor Gavin Newsom to enact or veto before the end of September.

TechRepublic reached out to OpenAI, Microsoft, and Adobe for comment.

SEE: Apple Joins Voluntary U.S. Government Commitment to AI Safety

Advocacy group criticises the bill

Microsoft and Adobe are part of the Coalition for Content Provenance and Authority, an industry group developing standards for the verification of AI-generated content. The group helped create C2PA metadata, a framework for embedding verifiable metadata into digital content that is used by DALL-E 3.

However, the BSA, an advocacy group for the software industry that includes Microsoft and Adobe, objected to AB-3211 in a letter to California’s Assembly Privacy and Consumer Protection Committee in April. It took issue with:

• The breadth of content that must be watermarked.
• The requirement to provide tools to verify watermarks.
• The “unworkable” notification and reporting system it proposes.

Since then, lawmakers have made several amendments to the bill, which appear to have appeased the companies’ concerns.

California juggling multiple AI-related bills

AB-3211 is separate from SB-1047, California’s controversial AI Act.

SB-1047 aims to prevent AI models from causing large-scale damage to humanity or financial losses by placing strict safety requirements on developers. Earlier this month, the California Appropriations Committee passed SB-1047, but it must still be approved by the state Assembly and Senate before it becomes law.

Much of Silicon Valley, including OpenAI, Meta, Google, and Hugging Face, publicly disparaged SB-1047. But earlier this week, Elon Musk took to X to express that he feels the bill should be passed. He posted: “For over 20 years, I have been an advocate for AI regulation, just as we regulate any product/technology that is a potential risk to the public.”

Two former OpenAI employees also wrote a letter to Newsom criticising their former employer for opposing SB-1047.

“OpenAI opposes even the extremely light-touch requirements in SB1047, most of which OpenAI claims to voluntarily commit to, raising questions about the strength of those commitments,” they wrote. OpenAI disagrees with the researchers’ “mischaracterization of [their] position,” per Business Insider

AB-3211 and SB-1047 are just two of at least 65 AI-related bills that lawmakers aimed to introduce in California this year, aiming to balance technological advancement with consumer protection.