CEI comments on Regulatory Reform on Artificial Intelligence

RE: Request for Information: Regulatory Reform on Artificial Intelligence

Docket No.: OSTP-TECH-2025-0067

The Competitive Enterprise Institute (CEI) appreciates the opportunity to comment on the Office of Science and Technology Policy’s (OSTP) Request for Information (RFI) regarding regulatory reform on artificial intelligence (AI). The RFI keenly recognizes that overregulation may unnecessarily or unintentionally hinder the progress, deployment, or adoption of beneficial AI technologies within the United States. CEI is a non-profit research and advocacy organization that focuses on regulatory policy from a free-market perspective. These comments represent the views of CEI as an organization; they do not purport to represent the views of any individual employee or of any donors.

Almost everyone now recognizes AI’s vast potential for economic efficiency gains, lifesaving medical breakthroughs, environmental benefits, improvement in education, and more. There is similar consensus around the importance of the U.S. winning the global race for AI dominance, ahead of foreign adversarial regimes, notably China for reasons of national security. But within our federal bureaucracy of hundreds of thousands of rules and guidance documents, estimated to produce a regulatory cost of $2.155 trillion in 2025, there are undoubtedly many government impediments to the domestic success of AI. These comments are not exhaustive and will focus primarily on antitrust policy.

However, it is worth reiterating that Congress has the responsibility to protect the free flow of interstate commerce. OSTP should encourage Congress to consider the constitutional benefits of a uniform federal framework. This would be a much-needed boon to innovations that benefit Americans, ensure U.S. dominance in the global AI race against China, and serve as a welcome assertion of constitutional congressional power.

There are currently approximately 1000 AI-related bills enacted or pending in states across the country. The compliance burden of that regulatory landscape has significant negative consequences for U.S. innovators trying to compete in a global marketplace. In a sense, we are asking our American AI innovators to compete with one arm tied behind their backs.

Beyond the practical problems with a patchwork of state rules, it is beyond question that large swaths of AI technologies obviously qualify as interstate commerce and, therefore, are a matter for federal instead of state authorities. Congress is well within its proper constitutional scope to prevent the strangling of AI technologies in state cradles of regulation. In fact, federal preemption of state AI regulations is likely the single best thing Congress could do to promote the full potential of AI in the U.S.

Antitrust

Legal scholarship has long discussed the imprecision and subjective application of antitrust law. This is particularly true for the foundational Sherman Act of 1890, which some recent commentary suggests is unconstitutional under the void-for-vagueness doctrine. Other scholarship addresses antitrust’s “unelaborated competition criterion,” describing it as a vague placeholder that causes indeterminacy and confusion. The Federal Trade Commission (FTC) and the Department of Justice’s (DOJ) Antitrust Division have issued guidance to provide more clarity and predictability in enforcement, “which can be changed or elaborated by the same judicial actors that gave it life.”

The White House took an important step in releasing its 2025 AI Action Plan. Under Pillar 1: Accelerate AI Innovation, the Plan recommends reviewing FTC actions from the previous administration that unduly burden AI innovation. The antitrust agenda of Lina Khan, particularly its targeting of the U.S. technology sector and firms at the cutting edge of AI, posed a significant threat to U.S. technological innovation, economic growth, and its competition with China, according to Joe Sullivan, former staff economist and special advisor at the White House Council of Economic Advisors. “[I]nnovators may also demur from technologically fruitful lines of inquiry for fear that it will run further afoul of Khan’s FTC, at least until the courts weigh in,” according to Sullivan.

At both the FTC and DOJ, the previous administration contributed to a lack of regulatory clarity relevant to the RFI’s Barrier 3. Specifically, this comment will focus on the lack of regulatory clarity created by the 2023 Merger Guidelines and the withdrawal of the 2000 Antitrust Guidelines for Collaborations Among Competitors.

Merger Guidelines

Mergers and acquisitions will play an important role in America winning the global AI race. According to Asheesh Agarwal, former attorney at the FTC and DOJ, “[w]hereas China uses intellectual-property theft and state industrial policy to advance its technologies, America’s innovation ecosystem relies on the robust flow of private capital to encourage investment, new entry, and patent filings.”
Agency leadership during 2021-2025 drafted and finalized the 2023 Merger Guidelines, with little input from Republican officials. The 2023 Guidelines were roundly criticized, and some hoped that there would be revision to the guidance. Despite the goal of providing clear guidance so that businesses can plan their mergers accordingly, the 2023 Merger Guidelines create heightened regulatory risk and legal uncertainty that can deter procompetitive mergers, according to Aurelian Portuese, professor and director of the GW Competition and Innovation Lab at George Washington University. “The 2023 MG will continue deterring venture capitalists from funding startups in the absence of clear exit options, and may deter large companies from acquiring small companies given the antitrust risks—both outcomes hurting the creation and expansion of small businesses,” Portuese wrote.

However, on February 18, 2025, FTC Chair Andrew Ferguson confirmed that the 2023 Merger Guidelines would go into effect and that the agencies would not be reworking the guidance drafted under the previous administration. Some called the decision an “opening blunder,” one that risks turning a flawed policy into a bipartisan consensus. The 2023 Merger Guidelines lowered the Herfindahl-Hirschman Index (HHI) thresholds for market concentration that trigger antitrust action, unsupported by case law or established economic theory. On the issue of lowered HHI thresholds, one of us wrote that
Evaluating mergers through this structural lens assumes both that concentration is present or escalating in the economy at large and that concentration always and only leads to harmful anticompetitive effects. Most detrimentally, it likely precludes the more nuanced assessment of a given merger’s impact on competition that the FTC had used over the preceding four decades. This shift will increase Type I errors in antitrust enforcement and rob or delay the benefits of competition-enhancing mergers to consumers.
The 2023 Merger Guidelines discourage beneficial AI mergers and acquisitions. These exit points are crucial for innovative startups to raise capital and increase scale. Further, according to Portuese, the 2023 Guidelines don’t provide clear guidance as to whether “business owners should make or buy the products they need to further compete against its rivals.”

The FTC and DOJ should review the 2023 Merger Guidelines to ensure they provide clear and predictable guidance that actively supports American innovation. A lack of clarity in the current antitrust approach creates unnecessary uncertainty for U.S. AI development, potentially hindering the nation’s capacity to maintain a competitive advantage in the global AI race.

Antitrust Guidelines for Collaborations Among Competitors

There is a lack of regulatory clarity as to how the antitrust agencies might view AI collaborations, because the previous administration withdrew prior guidelines without plans to issue new guidance. The FTC and DOJ, on December 11, 2024, jointly withdrew the 2000 Antitrust Guidelines for Collaborations Among Competitors (Collaboration Guidelines). Commissioner Melissa Holyoak, in her dissent, noted that the majority was doing so “without providing any replacement guidance, or even intimating plans for future replacement, leav[ing] business grasping in the dark.”
The drafting of new Collaboration Guidelines might provide needed clarity for certain AI collaborations, partnerships, and joint ventures. Shortly after withdrawing the 2000 Collaboration Guidelines, the FTC released its 6(b) study on Partnerships Between Cloud Service Providers and AI Developers. Commissioners Ferguson and Holyoak dissented to the inclusion of section 5 in the (6)b report, which was entitled “Areas to Watch Regarding Potential Implications of the AI Partnerships.” They noted that the report was quick and limited in scope, and Commissioner Ferguson said, “the limited, brief nature of the study should foreclose the drawing of broad conclusions about the AI industry and its future, or even about the partnerships themselves.”

Historically, the FTC has used comments, hearings, and public workshops to investigate evolving markets, which keeps its policies informed by current economic and technological realities. The FTC should replicate the process it utilized when considering the 2000 Collaboration Guidelines. Before soliciting comments on draft Collaboration Guidelines, the FTC should launch a notice of opportunity for comment and public hearing. The Commission should request public comments on the nature, purpose, competitive effects, and antitrust treatment of collaborations among businesses competitors. The comment period should be 90 days, following the process it used in the past. Further, the FTC should follow past practice by hosting a series of hearings and consider hosting a series of roundtable discussions.

This process should be employed now to address the implications of modern collaboration, particularly in the realm of artificial intelligence (AI) and its related safety and security implications. The antitrust agencies’ joint statement withdrawing the 2000 Guidelines, in a footnote, stipulated that the withdrawal did not affect the DOJ and FTC’s 2014 Antitrust Policy Statement on Sharing of Cybersecurity Information, which “make[s] it clear that they do not believe that antitrust is – or should be – a roadblock to legitimate cybersecurity information sharing.” The 2014 Policy Statement is limited, however, because it focuses on traditional cybersecurity threats. While calls for AI safety regulation are increasing globally, the current policy discussions sometimes fail to recognize that AI safety and AI security are deeply connected and should be addressed in tandem.

Changes in technology and the global marketplace led to both the initial creation and the eventual withdrawal of the Collaboration Guidelines. The previous administration’s actions may have created uncertainty for emerging AI collaborations and partnerships. This would be consistent with the priorities set forth in the White House’s AI Action Plan, which recommended the “[r]eview [of] all Federal Trade Commission (FTC) investigations commenced under the previous administration to ensure that they do not advance theories of liability that unduly burden AI innovation.”

Conclusion

Because antitrust law is often ambiguous, clear guidance is required to give industry the confidence to pursue innovative business decisions. While federal preemption of state AI regulation should be this administration’s foremost priority when it comes to regulatory reform for AI, the FTC and DOJ should also work to reconsider the unclear antitrust guidance fostered during the prior administration. CEI appreciates OSTP’s leadership in examining barriers to AI innovation and welcomes continued engagement on these issues.

Respectfully submitted,

Alex R. Reinauer
Research Fellow
Competitive Enterprise Institute
[email protected]

Jessica Melugin
Director of the Center for Technology & Innovation
Competitive Enterprise Institute
[email protected]