What: Independently deployed pricing algorithms can tacitly coordinate to keep prices above competitive levels, leaving the deploying company exposed even where antitrust law cannot reach.

Why it is important: Tacit algorithmic coordination leaves none of the visible traces that have so far drawn regulatory and consumer backlash, making behavioural audits of pricing systems a board-level responsibility.

In a Fortune column, François Candelon, Paul-Louis Andres and Augustin Manchon argue that pricing algorithms can sustain above-competitive prices while leaving none of the evidence antitrust enforcement relies on. After automated pricing spread among German gas stations in 2017, margins rose by about 38% in markets where two competing stations adopted it, and did not move where only one did. Experiments published in 2020 had already shown reinforcement-learning algorithms learning to price above competitive levels and to retaliate against price cuts.

The authors identify three patterns. In the "hub", competitors feed data into a shared provider, as in the RealPage case, where a proposed November 2025 DOJ settlement would restrict the data the software may use. In the "mirror", one firm predicts rivals' reactions, as the FTC alleges of Amazon's Project Nessie, which Amazon disputes. The "ghost", in which independent systems reach the same truce independently, is the hardest to detect or prosecute.

Because coordinated and competitive markets produce the same figures, the authors urge boards to map which systems can observe competitor prices, audit behaviour rather than results alone, run counterfactual tests that remove competitor signals, and document an auditable mandate for every pricing system.

IADS Notes: Regulatory attention to algorithmic pricing in retail has so far concentrated on conduct that is visible from the outside. The price-fixing allegations against Amazon involving Walmart and Home Depot (Bloomberg, April 2026) brought algorithmic pricing into antitrust scrutiny of market leaders, while New York's disclosure law and the industry's legal challenges against it pushed retailers to reassess data use and risk management (Forbes, February 2026). Adoption is nonetheless accelerating: Walmart's patents for machine-learning pricing systems, linked to electronic shelf labels across 4,600 US stores, coincided with state proposals to ban dynamic pricing (Financial Times, March 2026), and consumer backlash against opaque, continuously changing prices has made reputational damage a greater risk than forgone margin (The Robin Report, June 2026). In each of these cases, shoppers or regulators can see the harm. The tacit coordination described in this article produces no such signal, which moves the burden onto internal governance; an AI system registry, risk tiers and bounded autonomy (Harvard Business Review, July 2026) map closely onto the behavioural audits and auditable mandates the authors recommend.

Your pricing algorithm may have stopped competing without your knowledge