A growing push to use antitrust law as a tool to direct the path of artificial intelligence is premature and could backfire, according to a new commentary that urges enforcers to exercise restraint. The argument, published in response to calls for more aggressive intervention, warns that heavy-handed action in a fast-moving, research-driven market is more likely to snuff out innovation than protect competition.
The push to steer AI through antitrust
Several leading voices in competition law have argued that antitrust statutes should play a central role in deciding how AI develops. University of California at Berkeley Professor Prasad Krishnamurthy recently wrote that antitrust "can help determine whether AI will be shaped by competition that benefits all of its users or whether it's quietly steered by today's technology giants." He is not alone in suggesting that enforcement agencies do more to set the future course of the technology.
Krishnamurthy's proposal that public-private partnerships could help finance AI firms-addressing worries about acquisitions by large incumbents-is worth exploring. But the broader claim that AI development needs significant antitrust intervention now is not yet justified, the commentary contends.
Two reasons for caution
First, the fear that a handful of dominant companies will capture the technology is not supported by evidence. Critics point to a "paradox of AI competition," where thousands of AI startups depend on the very incumbents they might one day displace. Yet no tech firm, no matter its size, can realistically believe that AI will not upend its existing business. That is why the largest companies are racing to build their own AI tools and striking deals with the many startups entering the field. With so many firms competing, worrying that an existing online marketplace, search engine, or software publisher can block the shift is, the commentary argues, unrealistic.
Second, trying to steer AI's future without clear proof of anticompetitive harm in a well-defined market is more likely to do damage than good. Antitrust scholars debate whether over-eager enforcement that slows innovation is worse than lazy enforcement that lets dominant firms block rivals. But there is broad agreement that research- and technology-based markets change quickly and in unpredictable ways. Courts and enforcers should move with particular care when markets are evolving fast. That caution applies equally, if not more, to merger reviews.
In such cases, the right questions are: Do the parties really have market power, and can it last? Two decades ago, several tech firms looked poised to control e-book sales. Today, most of those names are answers to trivia questions. The counter-factual matters, too. Without the prospect of a deal, would the innovation at the center of a collaboration have been developed at all?
The cost of blocking exit paths
Krishnamurthy himself acknowledges that "many startups are created with the expectation of being acquired; restricting that exit pathway could reduce incentives to innovate." A barrier to exit is a barrier to entry. For many firms driving current advances, collaborations or acquisitions may be essential to their progress, and steering them away from those opportunities would be a mistake.
As the legal profession grapples with these issues, building a working knowledge of the technology is increasingly important. Resources like AI for Legal offer targeted training for professionals who need to understand how AI intersects with regulatory frameworks and commercial practice.
Why this matters for legal professionals
Antitrust lawyers and in-house counsel advising clients on AI-related deals will need to watch whether enforcement agencies take the bait. The debate highlights the tension between traditional merger-review caution and demands to act preemptively in a market that resists easy definition. The practical takeaway: dealmakers should prepare for more scrutiny, but the economic and factual underpinnings for intervention remain thin. Until enforcement agencies can point to durable market power in a clearly defined segment, the case for blocking or reshaping AI collaborations is weak.
Your membership also unlocks:
