HR8893-119

What This Bill Does

The Protecting Consumers from Deceptive AI Act would require the Director of the National Institute of Standards and Technology to establish task forces within 90 days. The task forces would support technical standards and guidelines for identifying content created or substantially modified by generative artificial intelligence. Their work would cover digital images, video, audio, and text.

For audio and visual material, the task forces would consider content-provenance metadata, watermarking, digital fingerprints, and other technical measures. Where technically feasible, they should seek cryptographically verifiable provenance and watermarks that are difficult to remove or conceal. They would explore interoperable standards that let social media and other online services identify, retain, interpret, and display those signals while accounting for circumvention and enforcement. For text, the work may include embedded provenance data, metadata, watermarks, digital fingerprints, or other measures.

The bill directs the task forces to inform private, consensus-based standards development where possible. NIST must include representatives from relevant Federal agencies, AI developers, standards organizations, detection-technology companies, social networks, messaging services, search engines, browser and mobile-operating-system developers, academia, civil society, privacy and human-rights advocates, news and image providers, creator and copyright-owner groups, labor organizations, AI testing experts, digital forensics specialists, cryptographers, and other participants NIST considers appropriate.

Each task force would have 270 days after its establishment to give NIST recommendations on its standards and guidelines. It must also report its activities to specified House and Senate committees within one year and annually for five additional years. The task forces must consider guidance for storing and displaying provenance information in a privacy-preserving way. That guidance could tell online services how to alert users when shared content contains provenance data, explain what the data disclose, and offer options to limit metadata with privacy implications.

The bill defines key technical terms but does not itself require an AI developer to watermark output, require a platform to display a label, make a NIST recommendation legally binding, create a civil penalty, or appropriate a specific amount. Adoption and enforcement would depend on later standards work, voluntary implementation, or separate legal authority.

Who Benefits and How

Consumers and online users could gain more reliable ways to recognize synthetic or substantially altered media if the resulting standards are adopted. News publishers, image providers, artists, writers, and other copyright owners could gain tools for authenticating origin and preserving attribution. Platforms and AI developers could gain interoperable technical expectations instead of incompatible provenance systems. Privacy and human-rights advocates receive a formal role in shaping how metadata are stored, displayed, and limited. Standards and detection organizations could gain new technical development work.

Who Bears the Burden and How

NIST staff must create and administer multiple task forces, recruit a broad membership, coordinate recommendations, and support annual congressional reports for six reporting points per task force. Participating AI companies, platforms, media organizations, labor groups, advocates, academics, and technical experts would bear meeting, research, testing, and drafting costs. Developers and online services that later adopt the standards could pay to add provenance systems, preserve metadata, label content, resist circumvention, and build privacy controls. Those private implementation costs are not mandatory under this bill alone.

Key Provisions

• Requires NIST to establish generative-AI content-provenance task forces within 90 days.
• Develops standards for metadata, watermarking, fingerprints, labeling, and interoperability.
• Includes government, industry, technical, media, labor, creator, privacy, and rights representatives.
• Requires recommendations within 270 days of each task force's establishment.
• Requires annual activity reports to Congress for an initial year and five additional years.
• Directs consideration of user notices and controls for privacy-sensitive provenance metadata.