$10 million fines await businesses whose pricing algorithms collude without anyone knowing

You are liable for what your algorithm does

If your business uses AI to monitor competitors and adjust prices in real time, you may already be exposed to cartel law, and the software vendor’s marketing brochure will not save you. That is the blunt finding of a new Bell Gully report published on 28 July 2026, the first substantive New Zealand legal analysis to map the Commerce Act directly onto AI pricing tools.

Bell Gully partner Glenn Shewan puts the liability question to bed early. “New Zealand’s competition rules apply whether decisions are made by people or algorithms,” he says. “Businesses remain responsible for the competitive outcomes produced by their AI tools.” Section 90 of the Act attributes the conduct of directors, employees and agents to the company itself, which means an algorithm’s output lands on the corporate balance sheet the same way an employee’s handshake with a competitor would.

The number that should focus board attention is the $10 million maximum fine for cartel conduct. That is not a theoretical ceiling. It is the figure directors need in front of them before they sign off on any AI pricing deployment.

Nobody has to intend it

The uncomfortable part is that collusion can happen without anyone in the business deciding to collude. The Bell Gully report maps several pathways. The clearest is using an algorithm as part of an explicit price-fixing agreement with competitors, which is obviously illegal. But the risk climbs from there into far greyer territory.

Agreeing with competitors to use the same pricing software can align prices across a market without a word being exchanged. A common software provider can act as a hub, with competing businesses effectively coordinating through it, so-called hub-and-spoke collusion. And most novel of all, AI systems can independently learn that parallel pricing produces better margins and quietly converge on coordinated behaviour with no human ever instructing them to.

That fourth pathway is the hardest to manage precisely because there is no smoking gun, no email, no meeting. The report warns AI tools can create competition risks “in ways that are harder to control and less visible than before.”

Overseas regulators are already prosecuting this

This is not a hypothetical drawn up by cautious lawyers. Shewan cites a US case in which landlords were caught using AI-enabled cartel conduct in the rental market. A White House report found as many as one in four rentals across the country may have been influenced by algorithmic pricing software, with pooled confidential data reducing genuinely independent decision-making across the industry.

Cases have now surfaced across five jurisdictions, the US, UK, European Union, Mexico and Brazil. Shewan’s read is that “international regulators are already taking action,” and that New Zealand businesses should expect increasing scrutiny given those trends.

The compliance gap most firms haven’t spotted

Here is the finding that should worry chief executives most. Liability may extend beyond the commercial teams who deploy pricing tools to the software engineers who build and configure them. The Commerce Commission’s own AI paper suggests exposure could reach the designers of the tool.

That is a genuine governance hole. Most competition compliance programmes are built around procurement and sales staff. The idea that a developer configuring a pricing model needs to understand cartel law is nowhere near standard practice in New Zealand. The Lawyer Magazine’s coverage of the report flags the same point, that technical teams sit well outside the usual compliance perimeter.

The regulatory net is tightening anyway

The report lands in an environment already moving one direction. The Fair Trading Amendment Bill introduced in May 2026 by Economic Growth Minister Nicola Willis and Commerce and Consumer Affairs Minister Cameron Brewer proposes lifting maximum Fair Trading Act penalties from $600,000 to $5 million, or three times the commercial gain. It is before select committee now.

The direction of travel was set back in 2023, when changes to the misuse of market power rules took effect. Then-Commerce Commission chair Dr John Small said the reforms would “stimulate greater competition by obliging the most powerful firms to have more regard to the competitive process.” The Commission has since been explicit that existing law already applies to algorithmic conduct.

The sectors most exposed are obvious once you know what to look for, retail and grocery, accommodation and short-stay rentals, fuel, construction materials, and online marketplaces, anywhere prices move fast in response to rivals and competitors share a common platform. The practical takeaway is unglamorous but urgent. Businesses need to understand how their pricing tools actually work, keep genuine human oversight over them, and extend compliance training to the engineers, not just the sales floor. The algorithm may be doing the pricing, but it is your company that answers for it.

Sources

• How AI could accidentally engage businesses in cartels – Newsroom (2026-07-30)
• Algorithmic collusion and AI cartels: risks under New Zealand’s Commerce Act – Bell Gully (2026-07-28)
• Bell Gully flags algorithmic collusion, AI-enabled cartel conduct as competition law challenges – The Lawyer Magazine (2026-07-28)
• Bill to crack down on misleading pricing introduced – 1News (2026-05-13)
• New Market Power Law Will Help Stamp Out Anti-competitive Behaviour – Scoop (2023-03-30)