California Passes AI Watermarking, Public Safety Risk Bills

California has passed two comprehensive bills providing AI content watermarking requirements and safety checks to prevent AI-triggered catastrophic events. Both bills were sent to Gov. Newsom for approval.

California's SB 942 mandates transparency in AI-generated content by requiring watermarks, according to Bloomberg Law.
The law targets a wide range of AI applications, including text, images, and videos.
The state also passed a critical piece of legislation for the development of AI models to prevent calamities such as weaponization and viral outbreaks.
Both bills have been sent to Gov. Newsom for approval.

The California legislature has approved SB 942, pioneering legislation requiring AI-generated content to carry identifiable watermarks, according to a Bloomberg Law report. The law, aimed at enhancing transparency and reducing misinformation, marks a significant regulatory step for AI technologies.

What the law says:

The new law mandates that all AI-generated content, such as text, images, and videos, must include a distinct watermark or identifier.
SB 942 places the responsibility on developers and companies using AI tools to comply with these transparency standards.
The legislation also includes provisions for penalties and enforcement.

Why it matters: The law comes amid recent AI advancements that have resulted in pictures, audio, and video that seem unprecedentedly realistic. The bill aims to ensure that consumers can distinguish between human-generated and AI-generated content, thus reducing the risk of misinformation.

The big picture: This is the latest in a series of bills passed by the California legislature tackling the development and use of artificial intelligence, signaling that the state is serious about regulating AI in the interest of public trust and safety.

AI Public Safety Risk Legislation

California also passed SB 1047, becoming the first state in the nation to enforce specific rules around how AI models are developed to prevent catastrophic risks like weapons development or virus outbreaks, Bloomberg said.

What the bill says: The proposed legislation is designed to hold makers of big AI models liable if their technology caused significant injury or carnage, requiring safety testing and a “kill switch” for programs.

Opposition: Tech groups have criticized both SB 942 as being technically unfeasible, and SB 1047, saying it can stifle innovation and virtually “kill” the AI sector.

What’s next: SB 942 has been sent to Gov. Gavin Newsom (D), who will have until the end of September to sign or veto the legislation. SB 1047 was also sent to Newsom, where it faces its biggest test, as the Governor has stated that he doesn’t want to overregulate the state’s dominant AI industry.