Legislation introduced in the House on Thursday aims to address concerns about the misuse of rapidly advancing technologies such as artificial intelligence (AI). The bipartisan bill would require the identification and labeling of online images, videos, and audio generated using AI. These measures aim to tackle the increasing prevalence of deepfakes, which can be difficult to differentiate from authentic content. Deepfakes created using AI have already been used to imitate President Joe Biden's voice, exploit celebrities' identities, and impersonate world leaders. The proposed rules would necessitate AI developers to incorporate digital watermarks or metadata to identify content created with their products, similar to how photo metadata records information like location and time. Platforms like TikTok, YouTube, and Facebook would then be required to label the content to alert users. The Federal Trade Commission, in consultation with the National Institute of Standards and Technology, would finalize the specifics of the legislation.

Violators of the proposed rules would face civil lawsuits. The bill, co-sponsored by Reps. Anna Eshoo (Democrat) and Neal Dunn (Republican), is part of several proposals seeking to address the risks associated with AI. If passed, the legislation would complement existing voluntary commitments by tech companies and an executive order on AI signed by President Biden. Although bipartisan support exists, it is unlikely that meaningful AI regulations will be passed in time for them to be effective before the 2024 elections. The bill has been commended by organizations and AI developers alike for its focus on embedding identifiers in AI content, which would help individuals gain control over the role of AI-generated content in society.