R Street Comment on FTC “Policy Statement Addressing AI Accuracy”
1411 K Street N.W.
Suite 900
Washington, D.C. 20005
July 30th, 2026
Federal Trade Commission
600 Pennsylvania Avenue NW,
Washington, DC 20580.
Re: FTC Seeks Public Comment on Policy Statement Addressing AI Accuracy
Docket: FTC-2026-0859
Submitted Electronically
This Document is Approved for Public Dissemination
Comments of the R Street Institute in the matter of the Federal Trade Commission’s Proposed Policy Statement Concerning The Suppression Of Accuracy In Artificial Intelligence Systems
By Spence Purnell and Adam Thierer
Thank you for providing the R Street Institute (R Street) with the opportunity to comment in response to the Federal Trade Commission’s (FTC) request for comment on the “Proposed Policy Statement Concerning The Suppression Of Accuracy In Artificial Intelligence Systems.”[1] R Street has published several reports and testimonies relevant to this proceeding in which we outline pro-innovation, pro-free speech policies for AI and digital platforms.
We believe that the FTC’s proposed policy statement potentially opens a real Pandora’s Box of problems in the way the government regulates how AI companies “may be manipulating the behavior of their AI systems contrary to reasonable consumer expectations for objectivity and accuracy.”[2] In this comment, we argue that the agency should focus on narrowly addressing state violations of interstate commerce where they exist without engaging in algorithmic speech policing at the federal level through efforts to determine “[AI system] outputs that are distorted by undisclosed ideological objectives.”
The Agency Is Right to Worry about the Growing State AI Regulatory Patchwork
The agency is correct to note that, “anti-innovation states’ recent efforts to regulate AI are concerning,” and that, “[t]he growing number of enacted and proposed state AI laws threatens to create a patchwork of regulatory regimes and compliance challenges for American companies.” According to a leading AI legislative tracking service, more than 1,800 AI-related laws are currently pending across the U.S.[3] The vast majority of these proposals are state laws. While not all regulatory in character, most are and many more are passing into law.
Most algorithmic commerce and speech activities are inherently interstate activities that cannot function efficiently under a fragmented system of fifty different state regulatory regimes.[4] State-by-state AI regulation threatens the free flow of interstate commerce and speech by forcing developers of nationwide AI services to comply with the most restrictive jurisdiction rather than designing products for a unified national market.[5] Companies generally cannot create fifty different versions of the same foundation model or chatbot. As a result, restrictive rules adopted by large states such as California or New York can effectively become national standards—the so-called “Sacramento effect”—raising compliance costs, discouraging smaller firms and startups, fragmenting markets, and impeding the development and movement of AI-enabled products and services.[6]
Regulatory fragmentation carries broader strategic consequences. Divergent and confusing state mandates could delay deployment of new AI capabilities, reduce economies of scale, complicate investment decisions, encourage defensive over-compliance, and make the United States less attractive for AI development at a time of intense international competition with China and other nations.[7]
While states have an important role in enforcing generally applicable laws against fraud and consumer harm, they should not each attempt to create separate regulatory systems governing the design, development, training, or deployment of AI models. Congress will need to establish a national framework that protects interstate algorithmic innovation, commerce, and speech to prevent the development of an “AI Articles of Confederation” policy model.[8] This is precisely the type of burden on interstate commerce that the Constitution assigns Congress authority to address through its Commerce Clause powers.
As the agency makes clear, the federal government, including the FTC itself, have an important role to play in addressing state and local regulations that implicate interstate commerce. While it is ultimately the responsibility of Congress to formulate a legislative framework for AI governance that addresses the federal-state balance of jurisdictional authority, the FTC has the ability to address restraints on commerce.
Sec. 5 Is Not a Blank Check for Algorithmic Speech-Policing
Unfortunately, this proceeding takes a different approach. The potential exists for this investigation of state laws and AI companies’ content policies to collide with important First Amendment values.
While the FTC possesses broad Sec. 5 authority to police unfair or deceptive acts or practices, when dealing with speech-related matters and government oversight of “algorithmic fairness,” those powers present real dangers of censorial over-reach through direct and indirect government interventions.
As the request correctly notes, some States like Colorado have already overstepped their bounds by “requiring AI systems to incorporate discriminatory ideology that prioritizes preferred demographic characteristics and outcomes over accurate and merit-based outputs.” However, the FTC incorrectly conflates this public policy requirement with the legal right of AI providers to program their outputs however they see fit.
The government policing ideological requirements is a completely different legal situation from AI providers themselves deciding to put ideological constraints on their systems. As the legal challenge to Colorado’s law notes, the design of such ideological systems in software, video games, and search engines has long been protected by the First Amendment.
Ironically, the FTC in one statement both identifies and further perpetuates the same mistake as Colorado’s law. It may seem like this request only requires that systems be honest to users about their systems, but in order to enforce this law, the government would have to compare stated ideologies and source code of each available LLM and compare those with outputs to verify that there is no “deception.” This enforcement would essentially amount to the government being able to decide if LLMs are sufficiently representing their ideological programming, a clear violation of First Amendment principles.
Even if it were not protected by the First Amendment, it is not considered “deceptive” that software programs or algorithmally-enabled systems have certain ideological dispositions without disclosing their source code. Video games, for example, may espouse certain political or social values without explicitly making those values known to consumers before the purchase. However, there are many ways of uncovering these values without a government mandate. There are, for example, research methods publicly available to communicate to users potential bias in AI systems.[9] Users can make informed decisions about which models to use if they perceive a certain type of bias. Video games similarly have professional reviews which can indicate to users bias or ideology before a purchase.[10]
Furthermore, the statement’s reliance on “consumer expectations” misuses that policy lever. The consumer expectations rationale is typically used in cases where there is an omission of information that should have existed.[11] Given that most LLMs already have disclosure statements, it seems unlikely that the FTC could explicitly prove that consumers were deceived about a product. That the primary focus of the agency’s investigation is disclosing “ideological ends” of LLM systems, it’s unclear exactly what type of ideological disclosure statement beyond what already exists would both satisfy the consumer expectations doctrine and be constitutional.[12] For these reasons, using deception and consumer expectations are widely inappropriate in the context of ideological design of LLMs.
Fundamentally, both video games and AI are software, both are created with code, and both share the same free speech protections. Having the government place special requirements on AI systems would be unfair, and could open the door for the government to more closely police speech products. Should video game creators have to state their political preferences to customers before release? Should the government review the video game and compare the content to those statements? This would clearly not be an enforcement of “deception” in content, but an abuse of the FTC’s power by policing ideological preferences or outcomes, which are clearly protected by the First Amendment. This power would undoubtedly be weaponized by future administrations to police speech to align with their respective political viewpoints.[13]
Global Ramifications of Over-Bearing American Speech Policies
The agency correctly notes that:
“the United States is the global standard-bearer for AI technologies. That American companies dominate every layer of the AI ecosystem should not be taken for granted. Geopolitical rivals are investing heavily in this sphere, hoping to inject their own companies and values into the marketplace. American dominance is thus about more than winning some abstract race. It is about ensuring that Americans can continue to feel that their values are being respected as they interact with, and benefit from, an AI-powered economy.”[14]
But this proceeding opens up the door to America undermining its own interests on the global stage by engaging in the very sort of ideological-based policing that the Trump administration has rightfully castigated European officials for in recent years.[15] The European Union and various member states have been engaged in over-zealous efforts to police online platforms for “hate speech” and other matters.
The administration has called out this behavior as “foreign censorship” as foreign governments improperly intervene in the ability of American companies to moderate speech as they see fit, as is their constitutional right.[16] Congressional lawmakers have likewise identified these problems. The House Judiciary Committee has held several meetings around the problems of foreign censorship and correctly concluded that the actions of foreign governments to pressure American companies was inappropriate.[17]
It would be incredibly ironic and misguided if this administration ended up doing much the same by aggressively policing algorithmic systems. It could undermine trust in U.S. companies and platforms globally.
[1] Federal Trade Commission, “FTC Seeks Public Comment on Policy Statement Addressing AI Accuracy,” July 1, 2026. https://www.ftc.gov/news-events/news/press-releases/2026/07/ftc-seeks-public-comment-policy-statement-addressing-ai-accuracy.
[2] Id.
[3] https://www.multistate.ai/artificial-intelligence-ai-legislation.
[4] Kevin Frazier and Adam Thierer, “Congress Should Lead On AI Policy, Not The States,” Law 360, Feb. 4, 2026. https://www.rstreet.org/commentary/congress-should-lead-on-ai-policy-not-the-states.
[5] Kevin Frazier and Adam Thierer, “No Single State Should Dictate National AI Policy,” Governing, Aug. 28, 2028. https://www.governing.com/artificial-intelligence/no-single-state-should-dictate-national-ai-policy.
[6] Adam Thierer and Logan Kolas, “On Artificial Intelligence Policy, It’s California Versus Congress,” Orange County Register, March 31, 2026. https://www.ocregister.com/2026/03/31/on-artificial-intelligence-policy-its-california-versus-congress.
[7] Adam Thierer Testimony at Hearing on “AI at a Crossroads: A Nationwide Strategy or Californication?” before the House Judiciary Subcommittee on Courts, Intellectual Property, Artificial Intelligence, and the Internet, Sept. 18, 2025. https://www.rstreet.org/outreach/adam-thierer-testimony-hearing-on-ai-at-a-crossroads-a-nationwide-strategy-or-californication.
[8] Kevin Frazier and Adam Thierer, “Congress Should Lead On AI Policy, Not The States,” Law 360, Feb. 4, 2026. https://www.rstreet.org/commentary/congress-should-lead-on-ai-policy-not-the-states.
[9] Kevin Schaul, “Are AI Chatbots Like ChatGPT Politically Biased? We Tested Them.,” The Washington Post, June 24, 2026. https://www.washingtonpost.com/technology/interactive/2026/06/24/are-ai-chatbots-like-chatgpt-politically-biased-we-tested-them/.
[10] Colin Moriarty, “The Escapist, Politics, and How Games Journalists Ruined Games Journalism — CLS Side Quest,” YouTube, July 30, 2018, 0:54. https://www.youtube.com/watch?v=6OeZb-yenr0&t=54s.
[11] Charlie Mitchell, “FTC’s Policy Statement on AI Draws Pushback from Free-Market Advocates,” Inside AI Policy, July 6, 2026. https://insideaipolicy.com/share/20135. Geoffrey A. Manne et al., “In the Matter of Nomi Technologies, Inc.: The Dark Side of the FTC’s Latest Feel-Good Case,” International Center for Law & Economics White Paper 2015-1, 2015, p. #. https://laweconcenter.org/images/articles/icle-nomi_white_paper.pdf.
[12] Federal Trade Commission, “FTC Seeks Public Comment on Policy Statement Addressing AI Accuracy,” July 1, 2026. https://www.ftc.gov/news-events/news/press-releases/2026/07/ftc-seeks-public-comment-policy-statement-addressing-ai-accuracy.
[13] Spence Purnell, “The Federal Trade Commission Should Not Be the Ideological Speech Police,” R Street Analysis, Feb, 12, 2026. https://www.google.com/url?q=https://www.rstreet.org/commentary/the-federal-trade-commission-should-not-be-the-ideological-speech-police/&sa=D&source=docs&ust=1783433160262563&usg=AOvVaw1bgVFjtE77322gIgKW1lZn
[14] Office of Public Affairs, “FTC Seeks Public Comment on Policy Statement Addressing AI Accuracy,” Federal Trade Commission, July 1, 2026. https://www.ftc.gov/news-events/news/press-releases/2026/07/ftc-seeks-public-comment-policy-statement-addressing-ai-accuracy.
[15] Jonathan Cannon and Adam Thierer, “The New Information Control,” R Street Analysis, Sept. 3, 2024. https://www.rstreet.org/commentary/the-new-information-control.
[16] Naomi Nix and Ian Duncan, “Inside the Trump administration’s bid to eliminate ‘censorship’ in Europe,” The Washington Post, March 20, 2026. https://www.washingtonpost.com/technology/2026/03/20/trump-eu-dsa-censorship/.
[17] Spence Purnell, “Concern Abroad, Censorship at Home: The Contradictions in U.S. Digital Policy,” R Street Institute, Sept. 10, 2025. https://www.rstreet.org/commentary/concern-abroad-censorship-at-home-the-contradictions-in-u-s-digital-policy/.