Two rival e-commerce sellers. Same category. Their owners have never met. But for three straight months, their prices moved in lockstep — one raises, the other follows within hours; one refuses to discount, the other holds the line too.
No meeting. No WeChat group. No emails. Just two independent AI pricing tools, both bought from the same SaaS vendor, both told to "maximize margin while tracking competitors."
Is that antitrust collusion?
Classical antitrust law is built around one word: agreement. To prove a horizontal cartel, enforcers had to show a "meeting of the minds" — humans talking to humans. Algorithmic pricing quietly removes that premise. Nobody talked. And yet the market outcome — parallel pricing, harm to consumers — looks identical to a cartel.
Three patterns compliance teams should be watching:
- Same-algorithm scenarios. When one vendor sells the "same brain" to every meaningful player in a category, that algorithm becomes the messenger. Traditional cartel logic still applies — the software just replaced the go-between.
- Hub-and-spoke via platforms. A platform sits in the middle, competitors sit on the spokes, and the platform's data interface lets everyone see everyone else's pricing intent. Several foreign courts have already recast this structure as a horizontal agreement.
- Self-learning tacit coordination. Two truly independent models, no shared data. They still converge on "if you don't discount, I won't either." No human can explain how the equilibrium was reached. This is where the classical agreement framework breaks down — and where regulators are actively rewriting doctrine.
China is not sitting this out. The 2022 amendment to the Anti-Monopoly Law expressly reaches conduct that leverages "data, algorithms, technology, capital advantages and platform rules." The 2021 Platform Economy Antitrust Guidelines already flagged algorithms as evidence of concerted conduct.
For companies using AI pricing tools — especially cross-border sellers on Amazon, Temu, Shein, MercadoLibre, Rakuten — three practical steps:
• Run an antitrust pre-review before any AI pricing tool goes live. Not just data privacy. Look at what inputs it uses, whose prices it references, and whether the output can produce parallel movement.
• Keep human-in-the-loop documentation. Even if the algorithm sets the number, prove a human reviewed it.
• Ask your SaaS vendor: is this same model licensed to competitors in my category? Is the data truly walled off? Put the answer in the contract.
Algorithms don't take the fall. Fines are calculated on revenue, and they land on the company — not on the model.
If your team is deploying AI pricing across multiple jurisdictions, this is a conversation worth having before, not after, the first regulator letter.
#AntitrustLaw #AICompliance #CrossBorderLaw #AlgorithmicCollusion #ChinaLaw
———
Mike Wu, China-licensed Attorney · Cross-border practice
4 languages (CN/EN/ES/JA)
Site: wzklaw.com/contact.php?source=linkedin · Book consultation: wzklaw.com/contact.php?source=linkedin