This browser is not actively supported anymore. For the best passle experience, we strongly recommend you upgrade your browser.
| 4 minute read

Carrot beats stick? The CMA publishes its initial report on AI foundation models

On 18 September 2023, the Competition and Markets Authority (CMA) published its initial report on AI foundation models (FMs) i.e. large-scale models that can be adapted to a wide range of tasks and operations. The Initial Report analyses the landscape of FMs – highlighting both the transformative change and significant benefits they could bring, as well as potential competition and consumer law (and other) issues that could arise. The CMA goes on to propose a set of seven principles that the CMA says “should guide the development and deployment of FMs”: (i) accountability; (ii) access; (iii) diversity; (iv) choice; (v) flexibility; (vi) fair dealing; and (vii) transparency.

The CMA intends to consult broadly over the coming months on both the principles and the report itself, following what it describes as a “collaborative and iterative” approach, which provides a framework that we expect to see the CMA use more widely when assessing future nascent and newly emerging technologies, especially under the forthcoming Strategic Market Status (SMS) regime. The CMA says it hopes its approach will help guide the market to more positive outcomes and realise the maximum potential of FMs.

Key messages from the CMA

The Initial Report follows the UK Government’s white paper on AI, published in March 2023 and the CMA’s response to the White Paper of June 2023, broadly supporting the Government’s approach of building on existing regulatory regimes whilst establishing a central coordination function for monitoring and support. In addition, in May 2023, the CMA launched its initial review of AI models. 

The Initial Report is the result of that review, in which the CMA engaged with over 70 stakeholders, including a range of FM developers, businesses deploying FMs, consumer and industry organisations and academics. It also considered publicly available information, including the latest AI research.

The report considers issues by reference to three themes: (i) competition in the development of FMs; (ii) the impact of FMs on competition in other markets; and (iii) consumer protection. For each theme the CMA identifies a range of possible market outcomes from “positive” to “more concerning” and seeks to identify factors that could push in either direction. For example, the CMA focuses on the key inputs required by FM developers, potential first-mover advantages through data feedback effects and the need for consumers to access accurate and reliable information about services they are using that rely upon FMs.

To address potential concerns, the CMA has proposed a set of seven principles, drawn from both its discussions with stakeholders and lessons learned from the evolution of other technology markets:

  1. Accountability: FM developers and deployers are accountable for outputs provided to consumers
  2. Access: ongoing ready access to key inputs, without unnecessary restrictions
  3. Diversity: sustained diversity of business models, including both open and closed
  4. Choice: sufficient choice for businesses so they can decide how to use FMs
  5. Flexibility: having the flexibility to switch and/or use multiple FMs according to need
  6. Fair dealing: no anti-competitive conduct including anti-competitive self-preferencing, tying or bundling
  7. Transparency: consumers and businesses are given information about the risks and limitations of FM-generated content so they can make informed choices.

CMA programme of engagement

The CMA plans to undertake a significant programme of engagement on the report and proposed principles, during which it will consult with a wide range of stakeholders, including consumer groups, leading FM developers, major deployers of FMs, innovators, challengers, new entrants, academic and other experts, across the UK and internationally. With UK Government hosting a global AI Safety Summit next month, this reflects a growing trend towards more international collaboration on AI issues. 

In addition, the CMA will work closely with the Government as well as other regulators, in the UK including via the Digital Regulator Cooperation Forum, and further afield with their international counterparts. The CMA’s current plan is to provide an update on its thinking on these principles in early 2024.

The CMA’s broader approach to digital markets

The cautious and considered approach taken by the CMA to the initial review reflected in the Initial Report provides important insights into how we might come to see the CMA approach future digital regulation in at least three key respects.

  • Known unknowns: the Initial Report spills as much ink on the “unknowns” as the “knowns”, with each section containing an extensive list of “uncertainties”. This intentionally cautious approach reflects what CMA officials have described as the importance of having “humility” over what is simply not known about new and emerging technologies and the potentially significant downsides from intervention in nascent markets.

     
  • Collaborate, participate, regulate: the CMA’s approach reflects the intention – made clear in CMA and government public discourse around the legislation that will give the CMA its new powers – that digital market regulation be approached as a collective endeavour, encouraging participation and partnership with industry stakeholders in the design and development of regulation. This acknowledges the information asymmetries between the CMA and industry stakeholders in these types of technologies and the importance of their participation in the regulatory discourse.

     
  • Guide, don’t force (unless necessary): the CMA makes clear its preference – especially in fast-evolving markets – to guide rather than seek to force markets to evolve in a way that benefits competition and consumers. The Initial Report explicitly cautions that overly burdensome regulation can make it difficult for competition and innovation to flourish (in particular, creating barriers to entry/expansion for smaller players), emphasising that any regulation needs to be “proportionate and targeted at identified risks”. This risk is front of mind for government and regulators - and the day after the CMA’s report was published, funding was announced for a pilot of a new AI and Digital Hub, to be run by DRCF member regulators (including, the CMA), which will provide businesses with tailored support with the AI regulatory framework.  But of course, behind every carrot is a stick: the CMA makes clear that it is “ready to intervene where necessary”.

Looking ahead

Overall, the Initial Report reflects a CMA keen to prove to observers in all quarters (especially those in Parliament) that it is a responsible authority that can be trusted with the enormous powers it will be granted under the Digital Markets, Competition and Consumers Bill. 

AI is an area where the CMA has the potential to make a significant mark on the global stage, with enhanced consumer powers and the new SMS regime giving the CMA power to potentially impose rules on SMS firms’ AI offerings – something the EU’s DMA does not explicitly cover – and the world will be watching. 

Subscribe to our Tech Insights blog for insights, updates and news from our experts - subscribe now!

Tags

foundation models, antitrust & foreign investment, ai