California law mandates invisible AI fingerprints to label synthetic media

A new California statute requires AI developers to embed hidden provenance data in generated images, video and audio, letting users verify whether content is human-made or synthetic.

California has enacted a pioneering law that obliges artificial-intelligence companies to embed invisible provenance information—so-called digital fingerprints—into every image, video or audio file they produce. Starting Aug. 9, users will be able to upload suspect media to verification services, like those run by the Coalition for Content Provenance and Authenticity, to determine if AI was involved. The legislation, drafted by state Sen. Josh Becker, Assemblymember Buffy Wicks and Rick Chavez Zbur, responds to concerns about AI-driven election misinformation, scams and deep-fake abuse.

It also sets a timeline: by Jan. 1, 2027, major tech platforms must offer tools that reveal AI-generated content in users' feeds, and by 2028, phones and cameras built in California must automatically store provenance data at the moment of capture. Federal lawmakers such as Adam Schiff and Ro Khanna are pursuing parallel measures to stop AI-fabricated political ads. Researchers like Stephen Sharp Queener caution that the platform mandates could significantly alter how information is consumed online. The law aligns with the EU’s AI Act, reflecting coordinated international efforts to increase transparency in synthetic media.