California has introduced new laws addressing various aspects of artificial intelligence (AI), focusing on risks, privacy, and user protection against deepfake technology. These laws, signed recently, will take effect in 2026.
The California Office of Emergency Services (CalOES) is mandated to collaborate with leading AI companies like OpenAI and Anthropic to assess potential threats posed by generative AI to the state's critical infrastructure.
Under the new legislation, AI service providers must disclose transparent information regarding the data sources used for training their systems, including the volume of data, its origins, and whether it contains legally protected content.
In response to growing concerns about the use of personal data, the laws expand existing privacy regulations to cover generative AI systems. These systems will be subject to strict privacy rules if they expose personal information such as names or biometric data.
California also emphasizes the importance of AI education, requiring the inclusion of 'AI literacy' in school curricula for subjects like math and science, ensuring students understand AI's functionality and ethical implications.
A unique definition of AI has been established, describing it as a system created by machines or engineering with varying levels of autonomy that generates results affecting physical or virtual environments.
In healthcare, the laws require providers to clarify when generative AI is used in patient communications. Additionally, regulations mandate oversight by licensed physicians for AI tools in healthcare services.
New rules also address the use of AI in automated calls, requiring disclosure when AI-generated voices are used to prevent public manipulation.
The legislation criminalizes the use of AI to create non-consensual explicit images and mandates social media platforms to implement reporting systems for deepfake content.
Furthermore, generative AI systems must digitally watermark their content to indicate AI origin, aiding the public in distinguishing AI-generated materials from real ones.
Political campaign regulations now require online platforms to remove or label AI-generated content that could mislead voters, and candidates can seek legal recourse for non-compliance.
The laws also protect actors' rights, requiring consent for using their likenesses in AI-generated content.
However, a controversial bill (SB 1047) was ultimately not passed, with Governor Newsom advocating for a broader regulatory approach to encompass smaller AI systems.
These laws mark a significant step for California in establishing new standards for AI regulation, aiming to protect citizens, promote transparency, and ensure ethical practices in the rapidly evolving AI sector.