Despite differences in jurisdiction and enforcement structure, the two laws share a clear architectural throughline.
August 2, 2026, was a notable convergence point in AI governance: This was the day both the EU AI Act and the California AI Transparency Act brought core transparency obligations into force. Although the two laws were developed independently and differ in scope, they converge on a single regulatory theme: synthetic media must be identifiable and traceable—and in many cases disclosed to users or downstream systems. Such transparency requirements are emerging first because they require the least and enable the most. They address the most acute risks, such as impersonation, deepfakes and other fraud, while being the most technically mature and immediately implementable levers available to regulators.
Under the EU AI Act (2024), transparency obligations are split based on operational roles. Providers must build systems that automatically announce themselves during direct interactions and embed machine-readable markers into synthetic outputs. The marking requirement applies to systems placed on the market from this date forward. Deployers must actively notify individuals when operating context-sensitive systems like biometric categorization, emotion recognition, creating deepfakes or generating public-interest AI text. This dual accountability framework is backed by significant penalties of up to €15 million or 3 percent of worldwide annual turnover as a ceiling per infringement.
California’s AI Transparency Act, enacted through SB 942 (2024) and amended by AB 853 (2025), took effect the same day and mirrors the EU’s focus on provenance. The law is currently written to apply to providers of generative AI systems that are publicly accessibly in the state and have over 1 million monthly users, with pending legislation, SB 1000, that would remove that user threshold. It requires the covered parties to embed latent, durable, machine-readable markers in AI-generated images, video and audio, and to make available a free public detection tool that reads those signals. Raw text is excluded here, implicitly acknowledging that reliable text provenance remains technically immature. Enforcement is restricted to government authorities at a fixed $5,000 per violation, with a separate violation for each day of noncompliance.
Despite differences in jurisdiction and enforcement structure, the two laws share a clear architectural throughline. Both require detectable provenance signals, qualified in each case by what is technically feasible. Both require that users be able to identify AI-generated content in key contexts. Both treat transparency as the necessary precursor to future safety and accountability obligations. Taken together, these regimes signal the emergence of a de facto global transparency baseline for synthetic media. Providers operating internationally are unlikely to maintain separate marking and detection systems for each jurisdiction; instead, they are expected to converge on unified provenance frameworks that satisfy both. This milestone marks the beginning of that harmonization and the beginning of a regulatory environment in which synthetic media is increasingly expected to be traceable by design.
For More Information
If you have any questions about this Alert, please contact Agatha H. Liu, Ph.D., any of the attorneys in our Artificial Intelligence Group or the attorney in the firm with whom you are regularly in contact.
Disclaimer: This Alert has been prepared and published for informational purposes only and is not offered, nor should be construed, as legal advice. For more information, please see the firm's full disclaimer.