UK competition regulator updates on AI foundation models

The head of the Competition and Markets Authority (CMA) in a recent speech has outlined the UK regulator's latest activity and thinking around artificial intelligence (AI) foundation models.

Sarah Cardell, chief executive of the CMA, recently spoke on AI foundation models and shared highlights from the UK competition regulator's update paper that set out competition risks for foundation model (FM) markets. The CMA also published a technical update report on 16 April.

Interlinked risks

The CMA takes the view that there are three interlinked risks that pose a threat to open, fair and effective competition in the development of the FM sector.

Firstly, firms controlling critical inputs for developing FMs may restrict access to them in order to shield themselves from competition;

Second, powerful incumbents could exploit their positions in consumer or business-facing markets to distort choice in FM services and restrict competition in deployment.

And thirdly, it points out that partnerships involving key players could reinforce existing positions of market power through the FM value chain.

'Real concerns'

Cardell made the point in her speech that, with a deeper understanding and having watched developments closely, the CMA has "real concerns".

She said: "The essential challenge we face is how to harness this immensely exciting technology for the benefit of all, while safeguarding against potential exploitation of market power and unintended consequences.

"We’re committed to applying the principles we have developed, and to using all legal powers at our disposal – now and in the future – to ensure that this transformational and structurally critical technology delivers on its promise."

AI principles

The CMA also updated the AI principles proposed in its initial report on foundation models. The principles are designed to ensure fair, open and effective competition for the benefits of consumers, businesses and wider society:  

• Access: ensuring ready access to inputs.
• Diversity: ensuring sustained diversity of business models and model types.
• Choice: ensuring sufficient choice for businesses and consumers so they can decide how to use foundation models;
• Transparency: giving consumers and business the right information about the risks and limitations of foundation models; and
• Accountability: ensuring that the developers and deployers of foundation models are accountable for their output.

When assessing the key risks against these principles, the CMA has indicated that it will take account of developments in FM-related markets when considering its enforcement priorities under the Digital Markets, Competition and Consumers Bill (DMCCB), which is currently nearing the end of its legislative passage through Parliament.

Consumer benefits and harms

The CMA's report also notes that the development of FMs may result in significant benefits and harms to consumers.

The potential benefits of higher quality, lower-priced and potentially more personalised products and services are contrasted with the possible extension of existing unfair practices and inadvertent harms caused by hallucinations or other flaws in the technology.

The DMCCB will give the CMA new powers to enforce consumer protection law and impose significant penalties for non-compliance, including fines of up to 10% of worldwide turnover.

Regulatory overload

The potential "chilling effects" of additional regulation and market intervention are also considered. The CMA identifies a risk of chilling effects on competition if interventions are so burdensome that only larger firms can comply. This creates the potential for raising barriers to entry for smaller firms or for those with disruptive business models.

With this in mind, the CMA endorses the House of Lords Communications and Digital Committee's recommendation that "market competition" should be an explicit policy objective of the UK government’s work on AI.

These concerns have been raised in conversations with fellow regulators in the UK through the Digital Regulation Cooperation Forum (DRCF). The CMA is also monitoring international developments such as the US White House Executive Order on Safe Secure and Trustworthy AI, the EU's AI Act and the Bletchley Declaration.

Following the release of the update paper, the Computer and Communications Industry Association, the international tech industry representative body, warned against intervening prematurely in the rapidly evolving market for AI services.

What next?

The CMA is using the paper as an opportunity to set out its future work on the impact of foundation models on competition and consumer protection.

It is planning to publish a further update paper in the autumn on foundation models, as well as a paper on AI accelerator chips and their role in the FM value chain. Separately, it is investigating competition in cloud services – which is how foundation models are typically accessed.

A joint statement with the UK Information Commissioner's Office is also planned on the interaction between competition, consumer protection and data protection in foundation models.

The CMA will continue its ongoing work as part of the DRCF, including joint research on consumers' understanding and use of FM services and its participation in the DRCF AI and Digital Hub pilot, launched on 22 April this year.

Separately, the CMA has been asked (alongside other major UK regulators) to report to the government by 30 April setting out its strategic approach to AI. This will be broader than foundation models although the work in this area will no doubt form an important part of the strategy.