Public Citizen
AI-generated text. This page was generated using artificial intelligence.
Public Citizen is a US nonprofit consumer-advocacy organization founded in 1971. Its AI-related work combines research, regulatory petitions, legislative advocacy and public campaigning. It connects AI governance with consumer and worker protection, democratic accountability and concern about concentrated corporate power.[1]
Organization and funding
Public Citizen's website serves two entities. Public Citizen, Inc., undertakes lobbying and organizing, while Public Citizen Foundation, Inc., focuses on research, public education and litigation and has more limited lobbying activity. Their work is presented under a shared name, but their distinct functions matter when attributing a formal filing or financial record.[2]
The organization says it receives no corporate or government money. Its methods include mobilizing supporters, investigating policy issues, petitioning agencies and litigating. Its broad public-interest agenda extends beyond AI; this article covers the interventions specifically connected to AI law and policy.[1]
Election deepfakes and the FEC
In its October 2023 comment on FEC proceeding REG 2023-02, Public Citizen argued that the federal prohibition on fraudulent misrepresentation of campaign authority could reach deceptive AI-generated campaign communications. Its theory focused on a candidate or agent falsely presenting another candidate as saying or doing something, thereby misrepresenting the identity of the speaker.[3]
The comment acknowledged the limited reach of this statutory route, particularly for communications by outside groups, and argued that broader legislation would be necessary. It also described the risk that synthetic media could help people dismiss genuine evidence as fabricated. This was an argument submitted to the commission, not a finding that every misleading campaign image violated existing law. The document records Public Citizen's proposed interpretation during the 2023 rulemaking debate.[3]
Broader recommendations on synthetic media
Robert Weissman's August 2023 submission to the President's Council of Advisors on Science and Technology advocated labeling AI-generated text, audio, images and video. It proposed content-provenance infrastructure, limits on deceptive electoral media, privacy protections and international agreements against governmental dissemination of deceptive synthetic material abroad.[4]
The submission also supported licensing for powerful generative systems and argued for a development and deployment pause, while recognizing the political difficulty of securing one. It warned that measures against automated public comments should not unnecessarily burden real people's participation. These were dated recommendations responding to perceived threats to trust and democracy, rather than a description of policies subsequently adopted. The submission treated disclosure and provenance as complementary, with neither alone solving every misuse problem.[4]
Research on lobbying
Public Citizen's February 2026 Generative Influence report analyzed federal lobbying disclosures from 2020 through 2025. Its search counted references to artificial intelligence, autonomous vehicles and systems, or data centers. It found 3,570 lobbyists with at least one matching activity in 2025; the report explains that these people could also lobby on other subjects.[5]
The methodology distinguishes individual lobbyists from lobbyist-client relationships, because one person may represent multiple clients. These counts measure disclosed activity under the report's search terms. They do not measure spending exclusively on AI, the direction of every client's position, or success in changing law. The report's definition also differs from the broad issue categories used on disclosure forms, limiting direct comparisons between rankings.[5]
In the accompanying release, Public Citizen's researchers called for enforceable sector-specific protections, greater technical capacity for regulators and protections for workers, children and communities. They opposed blanket federal preemption and deregulatory sandboxes. Those prescriptions express the organization's policy response to its findings; the lobbying counts alone do not prove that any particular legislative outcome was caused by corporate influence.[6]
Selected contributions
- PCAST submission on generative AI and democracy (August 2023).[4]
- FEC comment on deceptive campaign communications (October 2023).[3]
- Generative Influence analysis of lobbying disclosures (February 2026).[5]
Related articles
References
- ↑ 1.0 1.1 1.2 1.3 Public Citizen, About Us. Accessed September 12, 2026.
- ↑ 2.0 2.1 Public Citizen, Public Citizen and Public Citizen Foundation. Accessed September 12, 2026.
- ↑ 3.0 3.1 3.2 Public Citizen, Comment on FEC REG 2023-02, October 11, 2023. Accessed September 12, 2026.
- ↑ 4.0 4.1 4.2 Robert Weissman, Comments to PCAST Working Group on Generative AI, August 2, 2023. Accessed September 12, 2026.
- ↑ 5.0 5.1 5.2 Public Citizen, Generative Influence, February 2026. Accessed September 12, 2026.
- ↑ Public Citizen, One in Four Federal Lobbyists Now Work on AI, February 24, 2026. Accessed September 12, 2026.