New US Bill Aims to Increase Transparency Around AI-Generated Deepfakes
Lawmakers recently introduced draft legislation that could significantly boost transparency for synthetic media online. The “Content Origin Protection and Integrity from Edited and Deepfaked Media” (COPIED) Act, unveiled last week, looks to address the growing challenge of identifying authentic versus artificially manipulated content.
If passed, the COPIED Act would mandate online platforms to enable users to openly tag any AI-altered images, videos or other digital works with relevant details on origins. This “content provenance” would reportedly be machine-readable to allow for easy detection of synthetic versus genuine materials. Platforms must also ensure this provenance data cannot be deleted or modified without consent.
The bill targets any site with over $50 million annual revenue or 25 million monthly users that hosts user-generated materials. It gives two years to comply with tagging requirements for both copyrighted works and independently generated synthetic content. Removing or obscuring provenance from others’ copyrighted material would be prohibited without permission.
Policymakers cite the need for transparency as advanced AI grows more able to generate highly realistic fake audiovisuals and texts, sometimes dubbed “deepfakes.” The new rules aim to put creators back in control of authenticating their own works while educating consumers. Federal agencies will work on standardizing detection techniques, and competitions may reward the best provenance and labeling tools.
If passed, the COPIED Act could significantly strengthen oversight of rapidly evolving synthetic media online. By promoting better identification of authentic versus artificially doctored content sources, it has potential to curb harmful manipulation while still allowing innovative new forms of AI expression. The bill now awaits committee review.
