Protecting Consumers from Deceptive AI Act
- Sponsor
- Rep. Foushee, Valerie P. [D-NC-4]
- Committees
- Energy and Commerce Committee (primary)
- Last action
- Apr 23, 2026
Bottom line
The bill aims to combat deceptive AI content by establishing technical standards for identification and requiring mandatory disclosures for AI-generated media on both creation and distribution platforms.
What it actually does
This bill mandates the National Institute of Standards and Technology (NIST) to establish task forces for developing technical standards to identify content created by generative artificial intelligence, including watermarking and digital fingerprinting. It also requires generative AI application providers to incorporate machine-readable disclosures into AI-generated audio and visual content and obligates covered online platforms to display these disclosures to users, with enforcement by the Federal Trade Commission (FTC).
Proponents argue
Supporters argue that this legislation is essential to protect consumers from deceptive deepfakes, safeguard national security against AI-generated misinformation, and maintain an informed electorate by ensuring transparency about the origin of digital content. They emphasize the growing difficulty for individuals to distinguish real from AI-generated content, necessitating clear labeling.
Opponents contend
Critics may contend that the bill's technical mandates could be difficult to implement and keep pace with rapidly evolving AI technology, potentially stifling innovation or imposing significant compliance burdens on AI developers and online platforms. Concerns might also be raised about the feasibility of tamper-proofing disclosures and the potential for over-regulation.
The bill is relatively concise and clearly structured, allowing an informed reader to grasp its core provisions and implications within a reasonable timeframe.
Section 3(a)(1)
Establishment of NIST Task Forces for AI Content Identification Standards
This provision requires the Director of the National Institute of Standards and Technology (NIST) to establish task forces within 90 days of enactment. These task forces are charged with supporting the development of technical standards and guidelines for identifying AI-generated content, including content provenance metadata, watermarking, and digital fingerprinting for audio, visual, and text-based content. The goal is to make these identification methods cryptographically verifiable and difficult to remove.
Supporters argue
Supporters argue that establishing these task forces at NIST is a crucial first step to create a robust, technically sound foundation for identifying AI-generated content. This approach leverages NIST's expertise in standards development and promotes collaboration across diverse stakeholders to ensure effective and adaptable solutions.
Critics contend
Critics might contend that the rapid pace of AI development could render any developed standards quickly obsolete, or that the process of developing consensus-based standards through task forces may be too slow to address urgent threats. There could also be concerns about the feasibility of creating truly tamper-proof identification methods.
Tradeoffs
The provision balances the need for comprehensive, technically robust standards with the challenge of keeping pace with rapidly evolving AI technology and ensuring broad industry adoption. It also navigates the tension between government-led initiatives and private sector innovation in standards development.
Section 3(b)(1)
Mandatory Disclosure for AI-Generated Audio/Visual Content by Application Providers
This section requires providers of generative AI applications to ensure that any audio or visual content created or substantially modified by their applications includes a machine-readable disclosure acknowledging its AI origin. These providers must also implement measures to prevent the removal or tampering of such disclosures, collaborate with online platforms to facilitate identification, and offer users the ability to incorporate tamper-evident metadata about the AI application, model, date, and portion of content created by AI.
Section 3(b)(2)
Requirements for Covered Online Platforms to Display AI Disclosures
This provision mandates that covered online platforms (websites, internet applications, or mobile applications meeting specific revenue or user thresholds) must clearly and conspicuously display the AI-origin disclosure information to users for any audio or visual content accessed through their platform that incorporates such a disclosure. Furthermore, these platforms are prohibited from removing these disclosures or any associated metadata.
Section 3(b)(3) and (4)
FTC Rulemaking and Enforcement for AI Content Disclosures
This section directs the Federal Trade Commission (FTC) to promulgate regulations within two years of the Act's enactment to implement the disclosure requirements for generative AI content. In doing so, the FTC must consult with NIST and the established task forces. Any violation of these regulations will be treated as an unfair or deceptive act or practice under the Federal Trade Commission Act, granting the FTC full enforcement powers, including penalties and privileges.
Section 3(b)(6)
Safe Harbors for FTC-Approved Self-Regulatory Guidelines
This provision allows generative AI application providers and covered online platforms to satisfy the bill's requirements by adhering to self-regulatory guidelines that are approved by the Federal Trade Commission (FTC). The FTC is mandated to provide incentives for self-regulation, including deeming compliance with approved guidelines as compliance with the Act. The FTC must act on requests for guideline approval within 180 days.
The definition of 'covered online platform' includes specific thresholds: a website, internet application, or mobile application available in the U.S. that generates at least $50,000,000 in annual revenue OR had at least 25,000,000 monthly active users for not fewer than 3 of the preceding 12 months.
Section 3(c)(4)
Why it matters:This is standard legislative practice to define terms at the end of a bill. However, for a reader focused on the main duties, the specific thresholds that determine applicability could be easily overlooked.
Case for: Supporters argue these thresholds ensure the requirements apply to platforms with significant reach and resources, avoiding undue burden on small startups or niche services. This focuses regulatory efforts where the potential for widespread deception is greatest.
Case against: Critics might contend that even smaller platforms can host and disseminate deceptive AI content, and these thresholds create loopholes for emerging services or those with lower revenue but significant user engagement. This could lead to a fragmented information environment where AI-generated misinformation can still proliferate unchecked on non-covered platforms.
Estimated impact: Significantly narrows the scope of online platforms subject to the disclosure requirements, focusing on major players in the digital ecosystem.
The bill mandates that the NIST task forces consider issuing guidance for online service providers to store and display content provenance data in a privacy-preserving manner, including options for users to limit shared data. Similarly, the FTC is required to consider privacy concerns when promulgating regulations and approving self-regulatory guidelines.
Section 3(a)(5) and 3(b)(7)
Why it matters:This is a common legislative approach to integrate specific considerations into broader mandates. However, a casual reader might overlook the emphasis on privacy within the technical and regulatory directives.
Case for: Proponents argue that these provisions are essential to ensure that efforts to identify AI content do not inadvertently create new privacy risks for users or content creators. By explicitly requiring consideration of privacy-preserving methods and user control over data, the bill aims to build trust and prevent unintended surveillance or data exploitation.
Case against: Some might argue that achieving robust content provenance, which inherently involves tracking content origin and modifications, could be fundamentally at odds with absolute privacy. Critics might contend that balancing these two goals will be a significant technical and policy challenge, potentially leading to compromises on either front.
Estimated impact: Aims to mitigate potential privacy downsides of content identification technologies and policies, but the ultimate impact will depend on the specific guidance and regulations developed.