CMA focus on AI, collusion and competition law: What you need to know

As part of its wider focus on AI, the Competition and Markets Authority (CMA) has sent a clear signal to businesses that it is closely scrutinising whether algorithmic pricing tools and agentic AI are being used to facilitate collusion between competitors.

There have been two key recent developments in this space:

• On 2 March 2026, the CMA announced an investigation into the suspected sharing of competitively sensitive information between hotel chains Hilton, IHG and Marriott (the “Hotels Investigation”).  The hotels are alleged to have shared competitively sensitive information – including pricing data – through an algorithmic pricing tool operated by a third-party, CoStar. Notably, CoStar is also under investigation as operator of the price benchmarking tool.
• Shortly after announcing the Hotels Investigation, on 4 March 2026 the CMA published a blog on ‘AI and collusion: frontiers, opportunities and challenges’. The blog, which was published just two days after the announcement of the Hotels Investigation confirms the CMA’s view that AI tools should be used to support – rather than undermine – fair competition.

Double-edged technology: the pro- and anti-competitive potential of agentic AI

In its blog, the CMA acknowledged the transformational, pro-competitive potential of agentic AI for businesses, namely efficiency, speed and lowered costs, personalised offers, and increased market efficiency. AI can retrieve real-time data, seamlessly adjust offerings, and adapt to market conditions.

However, from the CMA’s perspective, these more advanced tech tools can also create competition risks. When rival businesses can access competitively sensitive information with such ease and speed through agentic AI tools, the uncertainty that they would typically have about one another’s practices, strategy and pricing may be reduced.

“Hub and spoke” collusion

CoStar’s AI platform being identified as a target of the Hotels Investigation aligns with a warning in the CMA’s blog against so-called “hub and spoke” collusion.

This is not a new theory of harm. As far back as the 2003 football shirts case, the OFT imposed penalties after finding that Umbro (upstream) had facilitated downstream collusion on retail prices between sports retailers.

In that respect, this is arguably just a case of well-known competition risks ‘moving with the times’. When businesses use the same algorithmic or data tool, it is important to consider where the information used to produce the results is drawn from. If data is used to provide bespoke pricing recommendations, the CMA is likely to have concerns if that data is mixed with information received from other businesses also using the tool.

The investigation into CoStar demonstrates that platform providers themselves may be liable for having facilitated any anti-competitive effects.

The CMA’s blog puts a specific onus on services which offer pricing algorithms, encouraging responsible behaviour and warning against:

1. Giving pricing recommendations to rival businesses based on confidential information drawn from each; and
2. Facilitating the exchange of competitively sensitive information between rivals.

The CMA’s expectations for businesses

The CMA encourages businesses to adopt a cautious approach when employing agentic AI tools, with the blog suggesting the following action is taken:

1. Train staff on the application of competition law to modern technology including any agentic AI being used;
2. Ensure that data from competing businesses is not pooled within shared platforms;
3. Actively scrutinise the application of agentic AI tools from a competition law perspective, “stress testing” prompts and models to avoid facilitating collusion; and
4. Report suspected anti-competitive behaviour to regulators like the CMA.

Interestingly, the CMA also took the opportunity in its blog to emphasise its own growing AI and technical capabilities, through the deployment of agentic AI tools to identify potential competition and consumer law infringements.

How we can help

TLT’s Digital Regulation team has market-leading experience advising on AI deployment. We work with businesses across the full lifecycle of AI deployment, from product design and procurement through to regulatory engagement. Our competition team has specific experience advising on pricing consultancy services and the risks associated with industry benchmarking.

Contributor: Tessa Lloyd

This publication is intended for general guidance and represents our understanding of the relevant law and practice as at April 2026. Specific advice should be sought for specific cases. For more information see our terms & conditions.