Assistant Attorney General Gail Slater Delivers Keynote at Fordham Competition Law Institute’s 52nd Annual Conference on International Antitrust Law and Policy

Remarks as Prepared for Delivery, "Winning the Race in Artificial Intelligence Through Real Competition"

Thank you for the kind introduction. Thanks to the organizers and to those who have traveled from near and far to exchange ideas at this conference. It's always a pleasure to see the depth of expertise and the breadth of interest in competition policy gathered in one place, and I am looking forward to hearing the debate. Of course, much like debate in general, debate about competition policy takes courage. As my grandmother liked to say, debate requires us to take a step back and recognize that God gave us two ears and one mouth for a reason. I wish us all the courage to listen to one another, because as C.S. Lewis wrote in the Screwtape Letters, "Courage is not simply one of the virtues, but the first of every virtue at the testing point."

Introduction

We find ourselves in the midst of a transformative time. Artificial intelligence is the new technological frontier. Its power promises to usher in a new era of technological, industrial, and informational change. Generative AI will revolutionize every sector of the economy and change how every business operates.

The tremendous potential of AI calls upon us to seize this moment with courage. President Trump and Vice President Vance have made it a priority to maintain AI dominance as part of the economic dynamism that is core to our success.[1] The success of the American economy is a testament to the fact that our free-market system works. Prioritizing innovation, supporting new technologies and startups, keeping the markets open for competition - these have long been the secret sauce of Silicon Valley and the cornerstones of American economic success. And these same core principles are what will help America lead the global race for AI.

The Trump-Vance Administration issued America's AI Action Plan at the White House in July, outlining the three pillars of our administration strategy: innovation, infrastructure, and international diplomacy and security.[2] Today, I'd like to focus on the first pillar, accelerating AI innovation. The antitrust laws have a lot to offer the innovation pillar, which is a core element of competition. Wherever there is a need for innovation, there is a need for competition, and this is where antitrust finds a home.

The good news is that the Antitrust Division is already fighting to keep AI markets competitive, and we are winning key battles in that arena. Earlier this month, the U.S. District Court for the District of Columbia issued its remedies decision in the Google Search case.[3] For more than a decade, Google illegally monopolized search markets, shutting out competitors, reducing innovation, and taking away choice from American consumers. In other words, the free market was frozen in place and innovation was therefore stymied. In the remedies decision, the district court held that "Google cannot use the same anticompetitive playbook" for GenAI as it did for search.[4] The judge ordered Google to share competitively critical data with qualified competitors, a group that the court made clear includes AI companies.[5] These data remedies, along with requirements that Google syndicate its search results to these competitors, are already setting the foundation for a more innovative and competitive AI industry.

Promoting AI Innovation Through Antitrust

As the Google Search example demonstrates, antitrust enforcement can help pave the way for innovation, dynamic competition, and development in AI. Let me explain how this works:

First, antitrust is, of course, the law of competition. A competitive environment is a prerequisite for innovation. As the AI Action Plan recognizes, it is critical that we "create the conditions where private-sector-led innovation can flourish."[6] Those conditions are those that incentivize and reward innovators and entrepreneurs. The story of American economic success is the story of innovative, disruptive Little Tech that grows up to become tomorrow's Big Tech.[7] Tomorrow's AI advances will be seeded by today's innovators.

Vice President Vance - who I had the pleasure of working for when he was Senator - put it best in Paris earlier this year: "This administration will not be the one to snuff out the start-ups and the grad students producing some of the most groundbreaking applications of artificial intelligence. Instead, our laws will keep Big Tech, Little Tech, and all other developers on a level playing field."[8] The antitrust laws are there to do just that.

Antitrust enforcement focuses on creating a level playing field for businesses big and small and ensuring that market incumbents do not unfairly hinder newcomers and startups. This is always important, but it is crucial where the technology is still developing rapidly. As the D.C. Circuit said in Microsoft - the last great monopolization case before the Google Search and Google Ad Tech cases - "[I]t would be inimical to the purpose of the Sherman Act to allow monopolists free reign to squash nascent, albeit unproven, competitors at will - particularly in industries marked by rapid technological advance and frequent paradigm shifts."[9] Antitrust has a framework that is uniquely equipped to help protect nascent competitors.[10]

Second, competition is also the key for opening up choices for consumers. Generative AI (GenAI) will have a profound impact on how we gather information, how we engage with media, and how our children learn. As Justice Thomas observed in 2021, today's digital platforms exert "unprecedented . . . concentrated control of so much speech in the hands of a few private parties."[11] The concentration of control created by AI may well surpass that. How AI is built - particularly at the foundation model level - can have the effect of imposing broad, implicit preferences over a wide swath of GenAI products. As the AI Action Plan correctly notes, this creates a significant risk of top-down ideological bias.[12]

How do we counter this risk of bias? The best way is to make sure that there is competition. Economic competition and viewpoint competition in the marketplace of ideas are closely intertwined. In a recent statement of interest, the Antitrust Division rejected the view that the antitrust laws play no part in protecting viewpoint competition in news markets.[13] Instead, antitrust can account for behavior that shuts out alternative, competing viewpoints. So too with AI. Ensuring that the AI markets are open to competition means that rivals with different approaches can compete on the merits. The availability of different AI systems and products can ensure that consumers have choices to pick from, safeguarding the free flow of information in our democracy. Different viewpoints can rise to the top in the marketplace of ideas.

Third, antitrust enforcement is a targeted tool suited for emerging fields like AI. In this regard, I like to compare the sledgehammer of regulation to the scalpel of antitrust. Premature regulation can be particularly harmful in incipient industries at the early stages of development because it imposes broad, ex ante rules, and these rules have the effect of limiting the direction of innovation across the entire industry.[14] From a competition standpoint, regulations can further entrench incumbent players, because only those incumbents can afford to pay the costs of compliance.[15] In fact, many powerful monopolists and incumbents lobby for regulations that raise barriers to entry and that ultimately harm market entrants and small businesses.[16]

By contrast, antitrust is law enforcement. It only intervenes ex post, in response to where illegal conduct has taken place, and it targets only specific conduct by specific bad actors on a case-by-case basis, rather than applying a one-size-fits-all approach to an entire industry. Such targeted enforcement strikes the right balance in emerging fields by finetuning the competitive conditions to make sure that the incentives are aligned for innovation to flourish. And if done correctly, timely, limited antitrust intervention can stave off the need for broad regulatory interference by letting the free market do what it does best.[17]

Ensuring the Competitiveness of AI Markets

So how does antitrust ensure that AI markets are competitive? To promote innovation, antitrust enforcement must focus on preventing exclusionary conduct over the resources that are needed to build competitive AI systems and products.

Building competitive models requires resources at every layer, from the infrastructure layer, to the foundation models, to the deployment of user-facing applications. The competitive dynamics at each layer of the AI stack and how they interrelate, with a particular eye toward exclusionary behavior that forecloses access to key inputs and distribution channels, are legitimate areas of antitrust inquiry.

Innovation is supported by the availability of key inputs, and so one important question is whether foundation models are open or closed. As our AI Action Plan rightly recognizes, open-source and open-weight AI models that are made available to developers "have unique value for innovation because startups can use them flexibly without being dependent on a closed model provider."[18] Of course, a truly open-source model must be one that is not unilaterally maintained by a single vendor that exerts unwarranted influence and impose restrictions. If done correctly, a truly open model can promote competition by making available a key piece of input into AI products. The ability to build on open foundation models to develop products removes a significant barrier to entry for startups and Little Tech.[19]

At another layer, an enormously consequential input for AI innovation is data. There is no question that AI intensifies the need for data as a critical input into progress in AI development. Enormous troves of quality data are needed to train, develop, and improve GenAI and LLMs. LLMs are pre-trained on large amounts of text, creating a base, or foundation model. The LLM's ability to predict the next item in a sequence depends on both the quality and the quantity of the data that it is trained on.

The centrality of data is not a new phenomenon in itself. Over the years, we have seen data grow in importance in the digital industries. Search is a good example of this. In finding Google liable, the district court noted that greater query volume yielded valuable user data that enabled Google to attain economies of scale and improve the quality of its search product.[20] Such a data advantage creates a flywheel effect: greater exposure to consumers leads to more data, which enables the product to be even better, which attracts more users, and so on. Companies that can amass and harness the power of big data gain an enormous advantage, while companies who lose out on data are shut out of the market.

These effects are not easy to overcome. That's why the Department of Justice and the plaintiff States asked for robust data sharing and syndication remedies in the Google Search case. And the district court agreed. The court recognized that "[s]earch index quality is critically important not only for traditional search engines, but also for emerging GenAI products,"[21] and that Google's data gave it a key competitive advantage over AI companies like ChatGPT.[22] And so the court imposed data remedies that require Google to provide qualified competitors access to its data and syndication of its search results, so that Google cannot continue to singularly exploit the data advantage it gained from years of illegal conduct. The data that is shared will give competitors - including emerging AI companies - a leg up to overcome the data gap and compete on a more even playing ground. The court was forward-looking, and it made it clear that "Google cannot be permitted to leverage its dominance in general search to the GenAI product space."[23]

While the scale advantages of data are not new, there is every indication that these dynamics will play out in a much bigger way with AI. The prospect of GenAI expands the competitive utility of data across every industry. Think about data on crop yields and local weather enabling more effective agriculture, or health data enabling more effective approaches in insurance. Now every industry is a data industry. As many have observed, data is the new oil. It fuels the engine of progress in every sector of the economy.

The finite amount of publicly available data also means that there will be more and more of a premium on developing proprietary sources of data. Access to high-volume, high-quality, proprietary data is a significant competitive advantage that could prove extremely difficult for competitors to surmount. Strong demands for data may drive the forces of consolidation. For example, vertical integration may become more attractive, especially in industries where downstream businesses may have access to valuable and sensitive data like healthcare data. We may also increasingly see the desire to acquire data, or to deprive rivals of data, play a role in driving transactions.[24]

Finally, I would be remiss not to say a few words about consumer data privacy. Having spent a large chunk of my career at the Federal Trade Commission and at various tech organizations, I am well versed in privacy issues and the intersection between competition and consumer protection. Data privacy is of course an important consideration, especially when data functions as an input into technology. But monopolists may also use privacy concerns as a pretext for gatekeeping data and refusing interoperability. We must closely scrutinize such claims.[25] Innovation around AI will likely be greater the more that AI products and services and their inputs are able to interoperate with each other.

Conclusion

In sum, there is a great deal that antitrust law can do to deliver the competition and innovation that will usher in the next golden era of AI. Races are won, and records are broken, through rivals pushing each other to do better-in other words, real competition. We will continue to fight for the competitive conditions needed to support American innovation and dominance in AI. And we will continue to support and play a role as needed in our Administration's AI Action Plan.

Thank you.