PROACTIV Artificial Intelligence Data Act of 2025
- Bill Number
- S. 2381
- Origin Chamber
- Senate
- Congress
- 119th Congress, Session 1
- Policy Area
- Science, Technology, Communications
- Status
- Introduced
- Latest Action
- 2025-07-22: Read twice and referred to the Committee on Commerce, Science, and Transportation.
- Last Updated
- 2025-12-02T12:03:45Z
AI-Generated Summary
Purpose
The legislation aims to protect children from online abuse by ensuring that datasets used to train artificial intelligence (AI) systems are free from child pornography. It requires the development of a voluntary framework to detect, remove, and report such illegal content, promoting safer AI development while providing legal protections for those who comply.
Key Provisions
- Definitions: Establishes clear terms, including:
- Artificial intelligence (AI): Refers to systems that perform tasks requiring human-like intelligence, as defined in existing federal law.
- AI developer: A person or entity that designs, codes, or produces AI systems for commercial use.
- AI deployer: A person or entity that integrates AI into products or services for commercial availability.
- AI user: A person using AI for non-personal, non-commercial purposes.
- Child pornography: Illegal visual depictions of minors engaged in sexually explicit conduct, as defined in federal criminal law (18 U.S.C. § 2256).
- Covered dataset: Data collected via automated tools (like web crawlers) specifically for training AI.
- Data collector: Entities that gather, clean, and organize large datasets for AI training.
- Director: The head of the National Institute of Standards and Technology (NIST), a federal agency focused on science and technology standards.
- Development of Framework: Within one year of enactment, the NIST Director must collaborate with agencies like the National Science Foundation (NSF), National Center for Missing and Exploited Children (NCMEC), and Department of Justice (DOJ) to create and publish a voluntary framework. This framework provides guidelines, best practices, and procedures for AI developers and data collectors to:
- Detect child pornography in covered datasets.
- Remove it from those datasets.
- Report it regularly to federal, state, or local law enforcement and NCMEC.
The framework applies only to AI developers and data collectors, not deployers or users.
- Stakeholder Engagement: The Director must seek input from universities, federal agencies, civil society groups, AI companies, federal labs, and others, including a public comment period.
- Research Support: The NSF Director, with other agencies, must fund research on innovative methods for detecting, removing, and reporting child pornography in AI datasets, including through NSF's technology innovation programs.
- Limited Liability Protections: AI developers and data collectors who follow the framework are immune from lawsuits related to their detection, removal, or reporting activities. Exceptions apply for intentional misconduct, acts with malice or gross negligence, reckless disregard for harm, or violations of federal child exploitation laws (18 U.S.C. § 2251). This does not alter existing reporting obligations under federal law (18 U.S.C. § 2258A).
Significant Changes to Existing Law
- Introduces the first federal framework specifically targeting child pornography in AI training data, which was not previously addressed in U.S. law.
- Adds targeted liability shields for AI-related activities, building on but not replacing general online service provider protections (e.g., under 18 U.S.C. § 2258A, which requires electronic service providers to report child pornography to NCMEC).
- Mandates interagency collaboration and research funding, expanding NIST's role in AI ethics beyond its current standards work.
Potential Impacts
- Government Agencies: NIST and NSF will gain new responsibilities for framework development and research funding, potentially increasing workloads and budgets. DOJ and NCMEC may see more reports, aiding child protection efforts but requiring resource allocation for investigations.
- Citizens: Enhances child safety by reducing the risk of AI systems perpetuating or generating child exploitation material through contaminated training data. AI users and the public may benefit from more ethical AI products.
- AI Industry: Encourages proactive data vetting, which could raise development costs but foster trust and compliance. No direct international relations impacts are outlined, though it may influence global AI standards indirectly.
- Broader Society: Could prevent the unintentional spread of child pornography via AI-generated content, supporting ongoing anti-exploitation initiatives.
Main Stakeholders Affected
- AI Developers and Data Collectors: Primary targets for compliance; gain liability protections but face new due diligence requirements.
- Government Entities: NIST (framework lead), NSF (research), DOJ (enforcement), NCMEC (reporting recipient), and law enforcement (investigations).
- Civil Society and Academia: Universities, nonprofits, and researchers provide input and conduct studies, influencing AI safety norms.
- AI Deployers and Users: Indirectly affected through cleaner AI systems, though not directly obligated.
- Children and Families: Ultimate beneficiaries via reduced online abuse risks.
Notable Legal, Constitutional, or Political Implications
- Legal: Provides a "safe harbor" from civil suits, similar to immunities in communications laws, but carves out exceptions for bad-faith actions to balance protections with accountability. Reinforces federal child protection statutes without creating new criminal penalties.
- Constitutional: No direct challenges anticipated; aligns with First Amendment limits on child pornography (which is unprotected speech) and government's interest in regulating technology for public safety.
- Political: Addresses growing concerns about AI ethics and child exploitation in a bipartisan manner (introduced by Sens. Cornyn and Kim), potentially setting a precedent for regulating AI data practices amid debates on tech accountability and innovation. As a voluntary framework, it avoids mandates that could spark industry resistance.
This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.
Sponsor
Cosponsors (2)
Sen. Kim, Andy [D-NJ], Sen. Budd, Ted [R-NC]
Recent Actions
- 2025-07-22: Read twice and referred to the Committee on Commerce, Science, and Transportation.
- 2025-07-22: Introduced in Senate
Bill Versions
- Preventing Recurring Online Abuse of Children Through Intentional Vetting of Artificial Intelligence Data Act of 2025 — issued 2025-07-22 — PDF (7 pages)