The Stanford Computational Antitrust Project announces the publication of “Generative AI Use by Competition Authorities: Why, Why Not, and What Might Be” by Richard May. The article appears in Volume 6 of Stanford Computational Antitrust (pp. 112-138).

Antitrust agencies can use AI to gather intelligence and to run their operations more efficiently. They can build tools in-house or rely on outside providers. The article examines what this adoption can deliver and what stands in its way. The potential gains are large. The risks are real. Some are particular to AI. Others come with any new technology.

The article argues that a top-down governance strategy matters most. Authorities need guardrails to manage the risks. They need scaling plans to turn early experiments into lasting capacity. Cooperation among authorities adds further value, since few agencies can address these problems alone. The article also looks beyond enforcement tooling. It asks whether AI might itself support pro-competitive intervention, for example through AI-based consumer agents acting on behalf of users. May treats this as a prospect rather than a recommendation. He argues it is too early to advocate such measures, while their potential merits further study.

Thibault Schrepel, founder of the Stanford Computational Antitrust Project, stated:

“Richard May moves the discussion past whether agencies should adopt AI toward how they should govern it. His focus on guardrails and scaling plans gives authorities a workable agenda. The closing reflection on AI-based consumer agents points to where the harder questions now sit.”