AI Antitrust Issues Checklist

As artificial intelligence (“AI”) transitions from a nascent technology into a central pillar of the global economy, public and private antitrust enforcement related to AI has likewise exploded. While the legal principles governing these complex systems remain fluid and far from settled, recent landmark agency investigations and new sector regulations demonstrate that legal exposure for certain AI practices is now an immediate reality. For any business currently developing, deploying, or commercializing AI solutions, understanding the following critical risk areas and executing clear, proactive compliance strategies is essential to navigating this high-stakes environment.

As explained in further detailed below, practices involving AI that could give rise to antitrust risk include:

• algorithmic collusion and improper information exchange
• integrating AI services into existing services with significant market position
• agreements among competitors on AI safety and standards
• exclusivity arrangements covering AI inputs and outputs
• mergers and acquisitions involving AI companies (including companies at different levels of the AI stack)
• restraints on AI labor markets
• compliance with new ex ante regimes for digital and AI markets.

AI Background and Global Regulatory Expansion

AI is now central to the technology industry and the global economy. Definitions vary, but AI is generally used as a catch-all for techniques including generative models, machine learning, neural networks, and complex algorithms used to perform specific or generalized tasks. Recent advances have moved AI from defined contexts (search, social media ranking, autonomous systems) into general-purpose applications such as conversational chatbots, code assistants, image and video generation, and agentic systems that plan and execute multi-step tasks. Large technology companies, established enterprises, and an expanding universe of startups now incorporate AI across their offerings.

In line with AI’s expansion in the global economy, authorities around the world are investigating and challenging AI-related conduct under the antitrust laws and, in some jurisdictions, new sector-wide regulations. In July 2024, the U.S. Department of Justice (“DOJ”), the U.S. Federal Trade Commission (“FTC”), the European Commission (“EC”), and the United Kingdom’s (“U.K.”) Competition and Markets Authority (“CMA”) issued a joint statement identifying risks from concentrated control of critical AI inputs, entrenchment of incumbents through partnerships and investments, and conduct that may foreclose nascent competition. Other authorities that have publicly addressed AI competition issues include the Japan Fair Trade Commission (“JFTC”), the Korea Fair Trade Commission (“KFTC”), the German Bundeskartellamt, the French Autorité de la Concurrence (“Autorité”), the Chinese State Administration for Market Regulation (“SAMR”), the Brazilian Conselho Administrativo de Defesa Econômica (“CADE”), and the Australian Competition and Consumer Commission (“ACCC”). Further, new ex ante regimes for digital markets and AI now impose affirmative compliance obligations on designated companies that build on top of traditional antitrust and competition enforcement. Private litigants are also a major source of AI antitrust risk. Class action and treble-damage claims involving algorithmic pricing, common pricing software vendors, no-poach arrangements, and exclusionary conduct have proliferated in the United States, with mixed results in early decisions. Companies developing or deploying AI should expect that enforcement risk will come from both regulators and private plaintiffs.

Importantly, regulators have identified concentration in the inputs to AI development as a competition concern of comparable importance to concentration in AI services themselves. The CMA’s final report on the U.K. cloud infrastructure services market, published in July 2025, found adverse effects on competition arising from concentration (with AWS and Microsoft each holding 30 percent to 40 percent of U.K. cloud spend), egress fees, technical interoperability barriers, and software-licensing practices. The European Commission opened market investigations in November 2025 to determine whether AWS and Microsoft Azure should be designated as gatekeepers under the Digital Markets Act (“DMA”) and whether the DMA’s obligations should be updated for the cloud sector.

Specific Issues for Consideration

Ensure that algorithms do not facilitate unlawful collusion or improper information exchange. AI algorithms and pricing tools remain a central enforcement focus across jurisdictions. The DOJ has brought claims against software vendors and their customers under a “hub-and-spoke” theory of algorithmic coordination, including in the RealPage Inc. (“RealPage”) litigation, where a November 2025 consent decree limits RealPage’s use of nonpublic competitor data and restricts certain design features that encourage acceptance of recommendations. The DOJ separately obtained consent decrees against several of the largest landlords that used the software, including Greystar and LivCor. Senior DOJ officials have made clear as recently as May 2026 that “[s]oftware cannot launder collusion,” and criminal antitrust enforcement may be applicable to certain algorithmic pricing schemes. Further, private class actions challenging the common use of pricing software are extensive, although plaintiffs have not been universally successful. For example, one appeals court held that common adoption of the same pricing vendor, without more, is insufficient to plead an antitrust violation. Additionally, several U.S. state legislatures have enacted or proposed statutes specifically targeting algorithmic rent-setting and pricing, and the EC’s 2023 Horizontal Guidelines and the JFTC’s recent generative AI study similarly treat algorithmic tacit coordination as potential concerted practice. Companies engaging in algorithmic pricing or recommendations can mitigate these risks by (1) avoiding nonpublic competitor data as inputs, (2) preserving independent pricing decisions and avoiding auto-accept or similar features that constrain users, and (3) implementing “compliance by design” controls and documentation around algorithm inputs, training data, and outputs.

Avoid exclusionary practices involving AI on existing services. Many technology companies are integrating AI into existing products in which they hold a strong incumbent position. Nevertheless, self-preferencing of AI features by digital incumbents is under scrutiny in the European Union (“EU”) under the DMA and in the U.K. under the Strategic Market Status (“SMS”) regime. The FTC is reportedly investigating Microsoft’s bundling of its productivity software with its AI, cloud, and security services. To mitigate these risks, companies should assess whether AI integrations may foreclose rivals or condition access to existing products on the use of complementary AI products, document the consumer-facing benefits of any integration, and consider whether interoperability or non-exclusive structures would suffice.

Ensure that agreements among competitors on AI standards and safety do not set prices or limit innovation. Cooperation among AI developers on safety, content standards, and risk mitigation remains a legitimate area for industry collaboration, and many such arrangements raise no antitrust concerns when properly scoped. However, horizontal agreements among competitors to fix prices or reduce output remain unlawful even in nascent industries, and regulators have indicated they will scrutinize joint commitments more closely where the participants are also the leading suppliers of foundation models or critical inputs. The EU AI Act imposes obligations on general-purpose AI models, in particular for models that pose systemic risk, so care should be taken to distinguish EU AI Act compliance, which is required by law, from voluntary standards that may extend beyond statutory requirements. Any cooperative arrangement should (1) be narrowly tailored to a legitimate safety, security, or interoperability purpose, (2) establish a floor rather than a ceiling for safety, and (3) not include exchanges of competitively sensitive information on price, output, customers, or commercial strategy.

Carefully consider exclusivity arrangements covering AI inputs, models, and distribution. Exclusivity requirements are not uncommon in agreements among AI developers, cloud service providers, chip suppliers, data licensors, and distribution partners. Regulators have looked closely at these arrangements where exclusivity covers an essential input or where it could limit AI companies’ ability to distribute their products or services. For example, the CMA’s review of the Microsoft/OpenAI partnership found that Microsoft’s status as OpenAI’s exclusive cloud provider afforded Microsoft material influence over OpenAI’s commercial strategy, but ultimately determined that the partnership did not constitute a merger due to certain concessions Microsoft made during the investigation. Accelerator chip markets are also under scrutiny: The Autorité has investigated Nvidia following a 2023 raid, SAMR opened an antitrust investigation in late 2024, and the DOJ has reportedly issued civil investigative demands focused on tying and exclusivity. Significant relationship-specific investments, scarce input procurement, and quality or security commitments remain standard justifications for exclusive dealing. Parties contemplating exclusivity in AI relationships should (1) document the procompetitive rationale, (2) consider less restrictive structures, such as preferred-provider or right-of-first-refusal arrangements, and (3) bound the scope and duration of exclusivity to what the relevant relationship-specific investment requires.

Mergers and acquisitions involving AI. Competition regulators continue to view AI as having significant potential to disrupt existing industries, and have signaled close scrutiny of AI deals, including minority investments and strategic partnerships with governance or commercial rights. The DOJ and FTC 2023 Merger Guidelines remain in effect in the United States and specifically articulate theories of harm relevant to AI transactions, including entrenchment of dominant positions, acquisitions of nascent or potential competitors, and mergers involving multi-sided platforms. The CMA, the EC, the Bundeskartellamt, the JFTC, the KFTC, and CADE have all expanded scrutiny of AI deals in recent reviews. Further, a growing share of transactions in the AI sector involves the “purchaser” hiring the founders and a substantial share of an AI startup’s employees, taking a license to the startup’s technology, and providing liquidity to investors, in place of acquiring equity or control. Notable examples include Microsoft/Inflection AI, Amazon/Adept, Google/Character.AI, and Meta/Scale AI. The CMA concluded that Microsoft/Inflection AI created a “relevant merger situation” within its jurisdiction, but cleared the transaction on substantive grounds; the Bundeskartellamt reached a similar conclusion on notifiability before declining jurisdiction for lack of local nexus; and the FTC has signaled that it will continue to scrutinize these structures for evasion of merger control. Businesses interested in acquiring an AI company, or in making a strategic investment in one, should (1) assess antitrust risks early, (2) plan for risks in transaction documentation and anticipate broader document requests, and (3) prepare to advocate for the deal across jurisdictions.

Avoid problematic restrictions relating to AI talent. The market for AI researchers and engineers is among the most competitive labor markets in the world, and antitrust enforcement in labor markets has remained a sustained priority across U.S. administrations and in major foreign jurisdictions. The DOJ has been active in this area going back to its 2010 enforcement actions against major technology companies for entering into non-solicitation agreements regarding specialized tech workers. More recently, the DOJ and FTC issued updated Antitrust Guidelines for Business Activities Affecting Workers in January 2025, which take the position that exchanges of competitively sensitive employment information, including through algorithmic intermediaries, can violate the antitrust laws even absent strict adherence to recommendations, and which remove safe-harbor language from the 2016 guidance for aggregated and anonymized exchanges. In general, labor market issues, including noncompetes, have been a priority of the FTC as reflected in the 2025 Labor Task Force directive issued in the first months of the Trump administration, which confirmed that non-solicit, no-poach, and non-compete agreements remain on the FTC’s radar. In line with this focus, the DOJ secured its first criminal wage-fixing trial conviction in 2025, and the Autorité fined three engineering and technology consulting firms approximately €30 million in June 2025 for no-poach agreements. Private no-poach and wage-fixing class actions are also active, with recent appellate decisions reviving claims at the pleading stage. No-poach, no-hire, non-solicitation, and wage-coordination arrangements involving AI talent (including those ancillary to investments, partnerships, or joint ventures) should be reviewed by counsel. Permissible ancillary restraints should be narrowly drafted and time-limited, and compensation surveys should follow established information-exchange principles.

Account for sector-specific ex ante regulation of digital and AI markets in compliance planning. Beyond traditional antitrust enforcement, companies developing or deploying AI now face affirmative compliance obligations under new regimes for digital and AI markets.

• In the EU, the DMA imposes a variety of obligations on designated gatekeepers’ core platform services, and the EC has identified AI and cloud as priority enforcement areas in its May 2026 DMA review; the EU AI Act, which regulates the development, marketing, and deployment of AI systems in the EU, applies in phases, with most remaining obligations applying from August 2026 and penalties of up to 7% of global turnover for prohibited-AI violations.
• The U.K. DMCCA entered into force in January 2025, and the CMA has designated multiple companies under the regime, with conduct requirements to follow, and has separately opened an investigation into Microsoft’s business software ecosystem.
• The JFTC’s Mobile Software Competition Act took full effect in December 2025, and the JFTC published its generative AI report in June 2025.

The KFTC, ACCC, and CADE have also opened market studies, investigations, or legislative work specific to AI and digital platforms. Companies should (1) maintain a current view of the jurisdictions in which they or their counterparties may be designated, (2) integrate ex ante compliance into product design rather than treating it as a post hoc overlay, (3) and recognize that compliance costs themselves may have financial implications, particularly for smaller developers.

This information is provided by Vinson & Elkins LLP for educational and informational purposes only and is not intended, nor should it be construed, as legal advice.