Felled by the Deregulatory Headwinds: DOJ’s Reversal on AI Divestiture in the Google Search Case

Kate Brennan is the Associate Director at the AI Now Institute. She is a former Google employee and a graduate of Yale Law School.

Recent headlines about the US Department of Justice’s case against Google’s search monopoly have rightly focused on the proposed divestiture of Chrome, a key structural remedy suggested by the DOJ. However, buried in Section Four of the DOJ’s initial remedy proposal was a bold and necessary requirement that Google divest any investment, holding, or interest in any rival AI company or technology product.

Why? One decisive way Big Tech companies achieve dominance in the AI market is by gaining control of competitive AI companies through partnership terms, investments, or acquisitions. This happens through direct investment (for example, Google’s investments in Anthropic totaling $3 billion), acquisitions of “everything but the company” (such as Google’s “aquihire” of Character.ai), or the purchase of smaller AI companies that evade formal merger review. Much of this happened within a permissive regulatory environment, applying conventional, restrictive interpretations to legal thresholds.

Because investments in AI companies are expensive and risky, only the largest companies with the biggest pockets can withstand the costs. As I wrote previously on Tech Policy Press, Google’s search monopoly profits are central to its ability to do this. Therefore, to fulfill one of the key goals of monopoly remedies—preventing Google from benefiting from its statutory violations—the proposed remedy targeted these investments.

Bold antitrust remedies—like the divestiture remedy initially recommended by the Department of Justice in the Google search monopoly case—are necessary to meet this current moment shaped by Big Tech dominance. Google, Microsoft, and Amazon control AI infrastructure from chips to cloud and deploy their vast resources to lock in users, influence regulation, and shape the political narrative.

Instead, strong anti-regulatory headwinds are taking shape across the globe and undermining the potential of competition authorities to tackle these market-dominating practices. Unfortunately, the DOJ’s recently revised proposal fits within this trend.

Last week, the DOJ removed the AI divestiture remedy from its revised proposed final judgment and instead requested Google to notify the DOJ of any future investments in rival AI companies, a significantly weaker provision. While the DOJ’s remedy proposal remains quite strong overall, this revision suggests closer alignment with the Trump Administration’s expressly accelerationist and deregulatory approach to AI development: promising to fuel AI data centers by any means necessary, platforming a $500 billion investment into domestic AI infrastructure, and allowing the owner of a private AI company bulldoze the federal workforce with its shoddy predictive technology.

The US is not alone. Just two days before the DOJ amended the Google divestiture provision, the UK’s Competition and Markets Authority announced it would end its probe into the Microsoft and OpenAI partnership. This came on the heels of the UK Labor Government announcing its “AI Opportunities Action Plan” intended to make the UK “the best place to start and scale an AI business.” In fact, the CMA’s recently released council minutes mark a decisively industrial policy turn, a markedly different tone than its earlier focus on safety.

Last June, the European Commission ended its probe into Microsoft and OpenAI’s partnership; in September, the Commission said it would not investigate the Microsoft and Inflection AI partnership agreement. At the Paris AI Action Summit in February, a string of announcements—from France’s €109 billion private investment in AI data centers to the European Union’s €200 billion investment to boost computing capacity—sent a clear message that any talk of strong antitrust enforcement will happen in lockstep with AI industry boosterism. Meanwhile, the European Commission and EU lawmakers are increasingly on the defensive after a string of criticisms from the Trump Administration about the EU’s investigations into American tech companies.

The DOJ’s revision on divestiture is one disappointing piece of an otherwise encouraging remedy package. The litany of remedies requested by the DOJ to end Google’s search monopoly will be a step towards ensuring that Google cannot use the fruits of its illegal search monopoly to achieve dominance in generative AI. For example, Google is prohibited from mandatory self-preferencing its own AI models (such as Gemini) on Android devices or other Google services. Google cannot force any publishers, websites, or content creators to exclusively provide Google with AI training data. Additionally, Google must also provide a way for all content publishers to opt-out of their data used to train or fine-tune any generative AI models or products. These are important to stop Google from abusing its position and forcing publishers to hand over their content for free or else risk deranking in search results.

But even if competition is restored in the search engine market tomorrow, the AI market will still be shaped by Big Tech power and dominance. Moves to create a truly competitive search ecosystem must work in tandem with meaningful enforcement actions targeting the broader AI ecosystem. Competition authorities around the globe must take a forward stance on investigations and enforcement actions to ensure a truly diverse and innovative tech ecosystem.