What is tacit collusion?
Tacit collusion is coordination between competitors achieved without any agreement or communication — each firm independently sustains supra-competitive prices because it expects rivals to follow increases and punish undercutting. Classical competition law struggles with it: without an agreement or concerted practice there is usually no infringement, however bad the outcome for consumers.
Why algorithms changed the debate
Pricing algorithms make the conditions for tacit collusion — market transparency, speed of reaction, credible retaliation — dramatically easier to satisfy. Learning agents can discover reward in mutual restraint and implement punishment strategies no human designed. Regulators’ answers are converging on three routes: recharacterising shared-vendor setups as hub-and-spoke concerted practices; attributing an algorithm’s conduct to the company deploying it; and, in Türkiye, proactive detection — the Competition Authority is building AI-based monitoring and has placed algorithmic pricing on its 2026 supervision agenda.
The compliance posture for algorithm users
Deployers should be able to show four things: no competitor-specific non-public data feeds the tool; no shared “market-level” brain with rivals via a common vendor; no retaliation or signalling logic in the objective function; and explainability — reconstructing from logs why a price was set. The defence “the algorithm did it” fails; the defence “here is the design, the data boundaries and the logs” works.
Is consciously matching a competitor’s public prices illegal?
No — conscious parallelism based on public information is lawful. The line is crossed with communication, shared private data, or mechanisms that stabilise coordination beyond independent adaptation.
Who is liable if a self-learning tool colludes?
The undertaking using it — knowledge is relevant to fines, not to attribution. Design-stage guardrails are the only real protection.
Related: economic moat.