AI is at a technological inflection point. The competition agencies around the world are trying to ensure they don’t fall behind in understanding and shaping this evolving landscape. In doing so, given the cross-border nature of AI related businesses and of the enforcement of antitrust laws, the joint statement released by the heads of the EU, UK and US competition agencies comes as no surprise but, at the same time, it has more meaning than meets the eye. It marks a shift from traditional ex post enforcement to a new ex ante application of competition laws.
On July 23, 2024, Lina M. Khan, FTC Chair; Jonathan Kanter, Assistant Attorney General with the U.S. Department of Justice; Sarah Cardell, Chief Executive Officer of the U.K. Competition and Markets Authority; and Margrethe Vestager, Executive Vice-President and Competition Commissioner for the European Commission, for the first time issued a joint statement outlining the specific risks AI poses to competition and the common principles they will apply in ensuring that the AI ecosystem remains competitive.
While this statement echoes many of the same themes that the agencies have expressed previously, it represents the first collective articulation of how these key jurisdictions will approach antitrust enforcement relating to AI foundation models and AI products.
Technological Inflection Point
The joint statement calls the proliferation of AI a technological inflection point that introduces new means of competing and thus requires vigilance against tactics that can undermine fair competition. Citing their learnings from digital markets, the agencies vow to address these risks “before they become entrenched or irreversible harms,” signaling their willingness to intervene where it may not be clear that a market outcome will inevitably be anticompetitive.
This is potentially a significant shift in the way antitrust law has been enforced to date, reflecting also a recent change in the tools available to the competition agencies to tackle perceived competition risks in digital markets. Historically, antitrust laws have been used to punish anti-competitive conduct ex post. However, the competition agencies felt that this approach proved to be too slow and unable to remedy competition harms after they had occurred.
The new approach underlining this joint statement advocates in favor of pro-competition interventions that aim to address potential risks to competition before AI markets tip irremediably with only a few or even only one provider/operator. This new ex ante approach is also possible thanks to the new regulatory tools now available to the competition agencies, such as the new UK Digital Markets Competition Consumer Act and the EU Digital Markets Act. While the U.S. Congress has not passed similar legislation, the FTC and DOJ have nevertheless intervened proactively – consistent with the directive in President Biden’s 2023 Executive Order regarding AI urging them to use the broad tools at their disposal – for example opening market inquiries and taking the unusual step of intervening in civil cases related to AI tools.
Key Risks to Competition
The enforcers identify three specific risks AI poses to competition, noting that their assessment will focus on how emerging AI models drive business incentives and ultimately behavior:
- Concentrated Control of Key Inputs: Because foundation model development depends heavily on critical inputs – namely specialized chips, substantial compute, large-scale data, and technical expertise – a small number of companies may be able to exploit existing or emerging bottlenecks across the AI stack.
- Entrenching or Extending Market Power: Large digital firms with existing market power may leverage it to protect against AI-driven disruption or advantage themselves through control of distribution channels needed to provide AI products and services to people and businesses.
- Partnerships Amplifying Risks: Collaborations and investments may be used by major firms in the AI space to suppress competitive threats and manipulate market outcomes in their favor at the expense of the public.
In assessing these three potential risks, the competition enforcers may be expected to assess not only the ability of firms to engage in conduct potentially giving rise to such risks but also their incentives (e.g. through partnerships, minority stakes or investments in developing AI technology).
Principles for Protecting Competition
While competition issues in AI will be fact-specific, three key principles will generally promote competition and foster innovation:
- Fair Dealing: Firms should avoid exclusionary tactics that can deepen their existing moats and stifle third-party investment and innovation.
- Interoperability: Ensuring that AI products and services can work together seamlessly will enhance competition and foster innovation, and notably the agencies will closely scrutinize claims that interoperability compromises privacy and security.
- Choice: Consumers and businesses should have access to diverse options, preventing mechanisms that lock them into specific products or services and restrict their ability to switch. This also means scrutinizing investments and partnerships to prevent incumbents from gaining undue influence or control.
Additional Competition and Consumer Risks
The deployment of AI involves other risks to competition itself, such as the potential for price fixing, unfair price discrimination, and sharing of sensitive information among competitors. The authorities will remain vigilant in monitoring these risks to ensure AI technologies are not used to undermine competition. Likewise, AI can exacerbate deceptive and unfair practices that harm consumers, including through use of consumer data to train models in a way that undermines privacy and security or divulges competitively sensitive information among businesses. Transparency is crucial to ensure that consumers are informed about when and how AI is used in products and services.
Broader Context
The competition agencies’ joint statement is at least one step in making good on previous promises to provide as much clarity as possible regarding their enforcement intentions as they intervene to shape market outcomes earlier than they might have historically in their efforts to avoid repeating past missteps with rapidly evolving technologies. Businesses developing and deploying foundation models and generative AI tools, of course, remain concerned that overly aggressive or unpredictable intervention will chill innovation in a still nascent and rapidly evolving space.
Anticipated Developments Against this backdrop, businesses will continue to urge against regulatory overreach while regulators will emphasize that moving too slowly will allow AI-related markets to tip and make it more difficult to correct competitive harms. Businesses should continue to look for ways to engage with regulators to educate them regarding how AI offerings work from a technical perspective and how markets operate in practice. As they have forecast earlier and potentially intrusive intervention than in the past, they have also indicated a willingness to engage with companies to develop their understanding of market realities.