In September, California Governor Gavin Newsom thought of 38 AI-related payments, together with the extremely controversial SB 1047, which the state Legislature despatched to him for remaining approval. He vetoed SB 1047 on Sunday, ending California’s controversial AI invoice geared toward stopping AI disasters, however signed greater than a dozen different AI payments this month. did. These payments search to handle essentially the most urgent points concerning synthetic intelligence. The whole lot from Allisk to deepfake nudes created by AI picture turbines to Hollywood studios creating AI clones of deceased performers.
“California, residence to most of the world’s largest AI firms, is leveraging these revolutionary applied sciences and learning the dangers they pose to assist handle urgent challenges.” Governor Newsom’s workplace made the announcement. press release.
Up to now, Governor Newsom has signed 18 AI payments, a few of that are essentially the most far-reaching legal guidelines on generative AI in america thus far. Here is what they do:
AI threat
On Sunday, Governor Newsom signed into regulation SB 896, which requires the California Division of Emergency Companies to conduct a threat evaluation of the potential threats posed by generative AI. CalOES will work with frontier mannequin firms reminiscent of OpenAI and Anthropic to research the potential threats of AI to crucial state infrastructure and threats that might lead to mass casualty incidents.
coaching information
One other regulation Newsom signed this month requires generative AI suppliers to reveal the info used to coach their AI techniques in paperwork printed on their web sites. AB 2013 Efficient in 2026, it’s going to require AI suppliers to determine the supply of their datasets, an outline of how the info shall be used, the variety of information factors within the set, whether or not it incorporates copyrighted or licensed information, and the info It’s obligatory to reveal the storage interval and so forth. Collected together with different requirements.
Privateness and AI techniques
Newsom additionally indicators AB1008 This makes it clear that California’s current privateness legal guidelines additionally apply to generative AI techniques. Which means that if an AI system like ChatGPT exposes somebody’s private info (identify, handle, biometric information), California’s current privateness legal guidelines restrict how firms can use that information for revenue. It is going to be executed.
schooling
Newsom indicators AB 2876 This month, it’s going to require the California State Board of Schooling to think about “AI literacy” in curriculum frameworks and supplies for math, science, and historical past. This implies California faculties might start instructing college students not solely the fundamentals of how synthetic intelligence works, but additionally the restrictions, implications, and moral concerns of utilizing the know-how.
One other new regulation is SB1288is asking on California’s superintendent to determine a working group to think about the best way to use AI in public college schooling.
Definition of AI
Newsom signed a invoice this month that creates a uniform definition of synthetic intelligence in California regulation. AB 2885 Synthetic intelligence is outlined as “synthetic intelligence or machines which have completely different ranges of autonomy and are capable of infer from the inputs they obtain the best way to produce outputs that have an effect on their bodily or digital surroundings, for express or implicit functions.” It states that it’s outlined as “based mostly system”. ”
well being care
One other invoice signed in September AB 3030This requires healthcare suppliers to reveal once they use generated AI to speak with sufferers, particularly when these messages embody affected person scientific info.
In the meantime, Newsom just lately signed SB1120This limits how well being care suppliers and medical insurance firms can automate companies. The regulation ensures {that a} certified doctor oversees using AI instruments in these settings.
AI robocall
Final Friday, Governor Newsom signed a invoice requiring robocalls to reveal whether or not they used an AI-generated voice. AB 2905 It’s geared toward stopping new circumstances of deepfake robocalls that resemble Joe Biden’s voice, which confused many citizens in New Hampshire earlier this 12 months.
deepfake porn
On Sunday, Newsom signed AB1831 This may increase the scope of current baby pornography legal guidelines to incorporate these produced by AI techniques.
Final week, Newsom signed two legal guidelines addressing the creation and distribution of deepfake nudes. SB926 It criminalizes this conduct and makes it unlawful to blackmail somebody utilizing an AI-generated nude picture that resembles oneself.
SB981The regulation, additionally handed Thursday, requires social media platforms to determine channels for customers to report deepfake nudes that appear to be them. The content material ought to then be quickly blocked whereas the platform investigates and, if confirmed, completely eliminated.
watermark
Additionally on Thursday, Newsom signed laws that may permit the general public to determine AI-generated content material. SB942 Requires broadly used generative AI techniques to reveal AI-generated content material of their content material origin information. For instance, each picture created by OpenAI’s Dall-E will want a small tag in its metadata to point that it was generated by AI.
Many AI firms are already doing this, and there are a number of free instruments that permit individuals to learn this provenance information and detect AI-generated content material.
election deep pretend
Earlier this week, California’s governor signed three legal guidelines cracking down on AI deepfakes that might affect elections.
One in every of California’s new legal guidelines, AB 2655requires giant on-line platforms like Fb and X to take away or label election-related AI deepfakes and create channels to report such content material. Candidates and elected officers can search injunctive reduction if giant on-line platforms fail to adjust to the regulation.
One other regulation, AB 2839is focusing on social media customers who put up or repost AI deepfakes that might mislead voters about upcoming elections. The regulation took impact instantly on Tuesday, and Newsom steered Elon Musk might be vulnerable to violating the regulation.
AI-generated political adverts now require full disclosure beneath a brand new California regulation. AB 2355. This implies President Trump could not be capable to get away with posting AI deepfakes of Taylor Swift supporting him on Fact Social (she helps Kamala Harris). The FCC has proposed comparable disclosure necessities on the nationwide degree and has already made robocalls utilizing AI-generated voices unlawful.
Actors and AI
Two legal guidelines Newsom signed earlier this month, championed by SAG-AFTRA, the nation’s largest display and broadcast actors union, create new requirements for California’s media business. AB 2602 Studios should get permission from actors earlier than creating AI-generated replicas of their voices and likenesses.
in the meantime, AB1836 Prohibits studios from creating digital replicas of deceased performers with out the consent of their estates (e.g., legally permitted replicas don’t exist in latest movies reminiscent of “Alien” and “Star Wars,” amongst others). (used within the film).
SB 1047 vetoed
Governor Newsom nonetheless has a number of AI-related payments to resolve on by the top of September. However SB 1047 was not amongst them, and the invoice was vetoed Sunday.
in a letter explaining his decisionNewsom stated SB 1047 focuses too narrowly on large-scale AI techniques and will “give the general public a false sense of safety.” California’s governor famous that small-scale AI fashions may be simply as harmful as these coated by SB 1047, and stated a extra versatile regulatory method is required.
Throughout a dialog with Salesforce CEO Marc Benioff earlier this month throughout the 2024 Dreamforce convention, Newsom could have tipped his hat to SB 1047 and the way he thinks about regulating the broader AI business.
“There’s one invoice that’s sort of substandard when it comes to public dialogue and consciousness, and that’s this SB 1047,” Newsom stated on stage this month. “What’s a demonstrable threat in AI? And what’s a hypothetical threat? We will’t resolve every thing. What can we resolve? It’s an method we use in all areas.”
Test again to this text for the most recent info on the AI legal guidelines that California governors signed and did not signal.
