California has officially implemented its AI Transparency Act, requiring artificial intelligence companies to embed indelible provenance data into generated images and videos. This move establishes a critical regulatory framework in the United States, stepping in as federal legislation remains stalled due to political disputes over whether national laws should override state-level regulations.

The Provenance Mandate and C2PA Standards

The law focuses on the technical ability to verify content authenticity. By requiring embedded metadata, California enables users to employ verification tools—such as those developed by the Coalition for Content Provenance and Authenticity—to distinguish between human-made and AI-generated media. This is specifically aimed at mitigating the risks of election misinformation and abusive deepfakes. Early compliance checks suggest that some industry leaders, including Midjourney, may face fines if they fail to implement these watermarking requirements.

A Phased Regulatory Roadmap

The legislation follows a strict timeline to ensure industry adaptation. Starting January 1, 2027, major online platforms must provide intuitive interfaces for users to inspect the provenance of content. The final phase arrives in 2028, when hardware manufactured in California, including smartphones and cameras, will be required to store provenance information at the moment of capture.

Global Alignment with the EU AI Act

California's law was developed in close coordination with European regulators, launching simultaneously with key provisions of the EU AI Act. While the EU's approach includes broader bans on high-risk AI systems that could threaten health or autonomy, California is focusing on the information economy. Given that Silicon Valley remains the global hub for AI development, this state law effectively sets a de facto national standard for the US industry.