S. 1396: Content Origin Protection and Integrity from Edited and Deepfaked Media Act of 2025

S. 1396, introduced on 2025-04-09, tries to build a traceability system for digital content at a time when deepfakes and AI-edited media are becoming harder to spot. The bill defines a “deepfake” as synthetic or synthetically-modified content that appears authentic to a reasonable person and creates a false understanding or impression. It also defines “content provenance information” as state-of-the-art, machine-readable information documenting the origin and history of digital content such as images, video, audio, or text.

The bill leans on the Commerce Department, especially the Under Secretary of Commerce for Standards and Technology, to stand up a public-private partnership and coordinate standards work with the Register of Copyrights and the Director of the USPTO. It also requires a public education campaign not later than 1 year after enactment covering synthetic content, deepfakes, watermarking, and content provenance. In parallel, grand challenges and prizes would be developed with DARPA and the National Science Foundation, signaling that the bill is not just about penalties but also about building workable technical systems.

The core compliance rules kick in 2 years after enactment. Companies that make commercial tools for creating synthetic or synthetically-modified content, or covered content, would have to let users include content provenance information and use reasonable security measures so that information stays machine-readable and is not easily removed or altered. The bill also makes it unlawful to knowingly remove, alter, tamper with, or disable provenance data in furtherance of an unfair or deceptive act, and it specifically bars covered platforms from removing or disabling that data except for security research purposes.

The reach of the platform rules is aimed at large services, not every website. A “covered platform” is a site or app available in the United States that has at least $50,000,000 in annual revenue or at least 25,000,000 monthly active users for at least 3 of the 12 months before a violation. The bill also goes after commercial AI training and generation practices: using covered content for a commercial purpose to train AI or generate synthetic content would be unlawful without express, informed consent and compliance with the copyright owner’s terms and compensation. Enforcement would come from the FTC, state attorneys general, and private lawsuits by owners of covered content, who would get 4 years from discovery of a violation to sue for injunctive relief, compensatory damages, and reasonable attorney’s fees.