Gov. Newsom vetoed CA SB 1047 however enacted no less than a dozen different payments regulating synthetic intelligence and associated applied sciences.
California Governor Newsom has vetoed SB 1047, the Protected and Safe Innovation for Frontier Artificial Intelligence Fashions Act, that will have amounted to the strictest regulations within the nation on the underlying massive language fashions (LLMs) that gasoline immediately’s rising generative AI applied sciences, comparable to OpenAI’s ChatGPT and Google Gemini.
Governor Newsom explained in his veto message that, “The invoice applies stringent requirements to even probably the most primary features […] I don’t consider that is one of the best method to defending the general public from actual threats posed by the know-how.”
If SB 1047 had gone into impact in California, authorities forms would have severely throttled non-public sector innovation in LLMs nationwide by arbitrarily limiting the lawful quantity of computing energy permitted to coach generative AI fashions. This authorities restriction on AI was proposed within the identify of public security, as an try to stop dangerous actors from unleashing “crucial harms” on society, together with using AI to develop a bioweapon, cyberattacks on crucial infrastructure ensuing from AI use, or “different grave harms to public security and safety.”
Regardless of good religion efforts from business and the AI analysis group to prioritize belief, security, and danger mitigation of their merchandise, SB 1047 would even have uncovered AI corporations to civil and legal legal responsibility for the unlawful conduct of third-occasion customers. If unable to show that their fashions don’t possess a “hazardous functionality” earlier than testing, LLM builders would have been focused by the regulatory state, amounting to a purposeful ban on open-supply AI fashions, comparable to Meta’s Llama 3, that deliberately share assets like developer toolkits with the broader open-supply group.
Though SB 1047 didn’t advance, Governor Newsom did enact at least a dozen other laws regulating varied features of AI in California. A few of these legal guidelines, together with AB 2839’s restrictions on using AI in election materials and political communications, have already prompted authorized challenges on constitutional grounds.
As a substitute of mimicking California’s pitfalls on AI regulation, states ought to comply with ALEC’s principled approach to AI regulation, utilizing present client safety and legal statutes on the books to handle demonstrable harms and hold criminals accountable for unlawful deepfake content material. It will permit policymakers to strike a correct regulatory stability that protects People, whereas responsibly accelerating the event of promising generative AI instruments on our shores.