Everyone is talking about AI. For businesses, the conversation is often about opportunity: greater efficiency, new products and a competitive edge. As businesses increasingly automate commercial decisions, for competition regulators the talk is more about risk: how those same tools could harm competition. Pricing tools and algorithms are attracting particular scrutiny worldwide, amid concerns that they could help competitors coordinate rather than compete. Today, the Commerce Commission has brought those risks into focus with new guidance on AI and Commerce Act and Fair Trading Act obligations, including guidance on algorithmic collusion.
The Commission recognises that AI and algorithmic tools can deliver benefits for consumers and businesses. Its focus is on uses that may harm competition, not the technology itself. But the Commission’s message to businesses is clear: “AI may be changing how decisions and outcomes are made, but it doesn't change who is responsible for them.” It may investigate where algorithms or AI systems facilitate cartel conduct or other anti-competitive outcomes, or result in a breach of consumer law protections.
AI changes how decisions are made, not who is responsible
Although MBIE’s proposed AI-specific clarification was not included in the proposed amendments to the Commerce Act before Parliament, the guidance makes the Commission’s position clear: the Act applies whether decisions are made by people, algorithms or AI systems and agents. The Commission’s message for businesses is that using AI does not shift responsibility for compliance, and businesses remain responsible for ensuring that their use of AI complies with existing provisions in the Act.
AI and algorithms can facilitate cartel conduct
The guidance identifies several ways AI and algorithmic tools may facilitate cartel conduct. These include software used to implement, monitor or enforce a cartel agreement, as well as risks arising through shared platforms, competitor information and autonomous pricing systems. For businesses choosing or using these tools, both the provider’s practices and the information used by the system matter.
A shared provider may facilitate coordination between competitors, for example if a common pricing platform or intermediary exchanges commercially sensitive information, or generates recommendations using competitors’ confidential pricing information. The Commission also highlights concerns where pricing algorithms are trained, calibrated or operated using competitors’ current or future commercially sensitive information. Autonomous pricing systems may learn to interact with competitors’ systems and produce coordinated outcomes without the businesses’ knowledge.
Ultimately, whether the cartel prohibition is engaged depends on the information involved, how it is used and evidence of a contract, arrangement or understanding between competitors.
The risks are not limited to businesses using AI tools. The guidance makes clear that those who design, manufacture or sell them may also face liability if they participate in cartel conduct or help put it into practice.
Overseas scrutiny is already leading to investigations and enforcement
The Commission’s guidance comes against a backdrop of increased overseas scrutiny of AI and algorithmic tools, with investigations and enforcement action particularly focused on pricing. Investigations have spanned sectors including travel, banking and pharmaceuticals. Enforcement has targeted businesses using software to implement price-fixing agreements, such as agreements not to undercut competitors. Software developers have also been fined for helping implement anti-competitive arrangements through their products. Other proceedings have examined allegations that shared pricing systems use pooled confidential competitor information to generate recommendations that help align prices. These concern the platform’s role in facilitating coordination, rather than simply carrying out an existing agreement.
For New Zealand businesses, these developments reinforce the importance of treating competition compliance as part of adopting and using AI.
Misuse of market power and merger prohibitions also apply to AI
The Commission’s guidance also discusses the prohibition on misuse of market power in the context of AI-enabled conduct. Businesses with substantial market power, including power accumulated through data, must not engage in conduct that has the purpose, effect or likely effect of substantially lessening competition.
The Commission identifies potential concerns where platforms use rankings or recommendations to favour their own products; essential digital infrastructure providers, such as cloud providers, require customers to take their AI services as part of a bundle; or businesses use AI-driven analytics for targeted pricing strategies that exclude rivals or new entrants. These practices are not automatically unlawful: substantial market power and the purpose, effect or likely effect of substantially lessening competition remain central.
The guidance also confirms that the prohibition on mergers and acquisitions likely to substantially lessen competition covers AI businesses and assets, including critical technologies, data and key AI capabilities. It highlights “killer acquisitions”, where a business acquires an innovative start-up to eliminate future competition and innovation.
AI compliance requires active governance and oversight
Commerce Act breaches can attract significant financial penalties, with individuals also facing up to seven years’ imprisonment for cartel conduct. Each separate breach may attract a penalty, making AI-related competition risks a material compliance issue.
The Commission is encouraging boards and senior leaders to treat AI as a governance issue. Its suggested guardrails are not exhaustive, but provide a basis for reviewing how tools are selected, used and monitored. They emphasise independent commercial decision-making: tools should not be designed, trained or configured to coordinate with competitors or respond to their conduct in a way that facilitates collusion.
Drawing on the Commission's guidance, the following checks provide a practical starting point:
- Keep competitive decisions independent: Check that pricing and other competitive decisions remain independent of competitors. Do not configure algorithms to coordinate with competitors or respond to their conduct in ways that facilitate collusion.
- Check data inputs: Do not train or calibrate algorithms using competitors' sensitive pricing, output or strategic information.
- Give reviewers the authority to intervene: Ensure people reviewing AI recommendations scrutinise the outputs and can override them.
- Test customer-facing outputs: Review AI-generated content for accuracy before publication. Test tools such as chatbots before launch and regularly during use.
- Ask suppliers how their tools work: Before adopting third-party pricing tools or AI platforms, establish what data they use, whether competitor information contributes to outputs and whether the supplier pools data across businesses.
- Update training and escalation processes: Cover algorithmic and AI-related risks in compliance training, and make clear who staff should contact if they identify unexpected AI behaviour.
- Keep a record of key decisions: Document algorithm design, data sources, risk assessments, audit findings and decisions to adopt or modify AI tools.
Businesses concerned about potential cartel conduct should seek legal advice and consider early engagement with the Commission.
Fair trading obligations also apply
The Commission’s guidance also addresses AI-related risks under the Fair Trading Act, covering how businesses use AI to advertise, communicate with customers and generate content.
AI-generated images and text, chatbot responses, fake endorsements and reviews can all mislead consumers. Businesses remain responsible for the content they use and must have reasonable grounds for claims when they make them. The Fair Trading Act guidance cautions against relying solely on AI-generated material as evidence and recommends reviewing content before publication and regularly testing customer-facing tools for accuracy.
Fair Trading Act breaches can attract significant fines, including up to $600,000 for businesses. Each separate breach may attract a penalty, making AI-generated content and customer communications a material compliance risk.
Our takeaway
The takeaway is not to avoid AI, but to manage its use with the same care as other commercial decisions. Businesses should understand how their tools influence pricing and other competitive decisions, what information they use, what their AI tools are saying and how they influence consumers and when human intervention may be needed. AI may change how decisions are made, but outsourcing a task does not outsource responsibility for the outcome.
As the technology evolves, so may the competition and consumer law risks. Businesses should stay up to date with regulatory guidance and enforcement in New Zealand and overseas, review their approach as tools and uses change, and seek legal advice early to manage competition risks and reduce exposure to enforcement action.
Our competition and consumer law and technology teams can help you assess how the guidance applies to your AI tools, third-party arrangements and compliance programmes.
For the full guidance, see the Commerce Commission’s AI guidance for businesses.
This article was co-authored by Damian Bachor, a Senior Solicitor in our Competition team.