The UK’s Competition and Markets Authority (CMA) is examining how AI foundation models are developing and how their use could affect competition and consumers. The review covers models such as large language models and generative AI systems, which can support products ranging from chatbots to AI art platforms.
The CMA says the work is intended to clarify the conditions and principles that could guide the development and use of these models. Stakeholders have until June 2 to submit responses, and the regulator proposed publishing its review in early September.
What the CMA wants to understand
The review focuses on questions within the CMA’s remit: what implications the development of AI foundation models may have for competition and consumer protection. The regulator says other concerns, including safety, security, copyright, privacy and human rights, are also being considered by government or other regulators.
That focus reflects the CMA’s core mission to support open, competitive markets. Its review is not presented as a complete assessment of every risk associated with AI. Instead, it is an initial effort to assess the market and consumer issues the authority is best placed to address.
The CMA describes foundation models as having the potential to transform activities for people and businesses. Its stated aim is to help ensure that innovation benefits consumers, businesses and the UK economy, while people remain protected from issues such as false or misleading information.
Principles for developing AI
The government has asked regulators, including the CMA, to consider how innovative AI development and deployment can be supported against five principles. The CMA listed them as:
- Safety, security and robustness
- Appropriate transparency and explainability
- Fairness
- Accountability and governance
- Contestability and redress
The principles frame the government’s approach to AI oversight. According to the article, a white paper published in March signalled ministers’ preference to avoid creating bespoke rules or new oversight bodies for AI at this stage. Instead, existing regulators are expected to provide guidance suited to their areas of responsibility.
The CMA says its review follows the white paper’s call for regulators to conduct detailed risk analysis. That work could help them decide whether enforcement is appropriate for dangerous, unfair or unaccountable AI applications, using their existing powers.
Why the review could matter to AI companies
The CMA’s current review is an assessment, but the article also points to possible future changes in the regulator’s role. Plans to give the CMA additional powers over Big Tech are set to move forward, with the Digital Markets Unit expected to gain legislative powers in the coming years.
Those powers are expected to let the unit apply proactive rules tailored to platforms designated as having “strategic market status” (SMS). The article suggests that powerful foundation model providers could potentially face such a designation in the future. That is presented as speculation, rather than a decision made by the CMA.
If that kind of oversight eventually applies, rules could address how designated companies operate in relation to rivals and consumers in the UK market. The initial review may therefore help build the regulator’s understanding of a fast-changing field, even as the future scope of its powers remains unsettled.
Other regulators are looking at AI
The CMA is not the only UK regulator paying attention to generative AI. The Information Commissioner’s Office (ICO), the country’s data protection watchdog, has also been tasked with considering AI under the government’s approach of context-specific guidance.
In a blog post last month, Stephen Almond, the ICO’s executive director of regulatory risk, advised organisations developing or using generative AI to consider data protection obligations from the outset. The article reports that he highlighted a data protection by design and by default approach and said that handling personal data brings legal obligations.
The article contrasts the UK’s reliance on existing regulators and guidance with the European Union’s work on a fixed set of AI rules. EU lawmakers were still negotiating the final text of their incoming rulebook, including how foundation models or general purpose AI should fit within its risk-based framework.
The exact rules and responsibilities remain under discussion. The CMA’s review adds a UK competition and consumer protection assessment to a wider regulatory conversation about how generative AI should develop and be used.