Don’t forget about competition just because it’s not a killer robot: CMA responds to DCMS’s paper on AI regulation

This summer the Department of Digital, Culture, Media and Sport (DCMS) proposed a "light touch” approach to AI regulation, aiming for sectoral and principles-based regulation rather than the central AI specific legislation being developed in the EU. In response the CMA has expressed support for DCMS’s approach - recognising the developing nature of AI and the importance of AI innovation as a driver for growth - but emphasising the need for an enforcement backstop to deal with the more egregious harms.

Potential harms of AI

The CMA has long had AI on its radar – and it continues to monitor AI developments through what the CMA terms its proactive market monitoring and horizon scanning capabilities. Its first economic working paper on the potential harms of AI in competition law was published in 2018, focusing on AI as a tool to further coordination between competitors. 

The CMA built on its earlier studies through a further research paper on algorithms, published in 2021 and, most recently, through the Digital Regulation Co-operation Forum (DRCF) to produce a research paper on the benefits and harms of algorithms, last updated at the end of September 2022.

The CMA notes that while competition harms may not have the immediacy of health and safety risks or harm to the fundamental rights of natural persons associated with AI, they have the potential to have a diffuse impact across many people over a long period of time. 

The CMA encourages more emphasis on long-term structural risks to competition as part of any regulatory framework and calls for such longer-term risks to be factored into any risk-based principles. The CMA highlights that such “structural” risks could arise through market power going unchecked, noting specifically:

1. The potential for concentration of key inputs to AI supply chains, such as chips, data and computational resources; and
   
    
2. The ability of powerful firms to act as quasi-regulators through implementation of rules and systems, including using AI.  

Cross-sector collaboration

The CMA also emphasises the need for cross sector collaboration to balance issues such as competition and privacy and, unsurprisingly, regulatory powers to ensure effective (ex ante) intervention. The CMA highlights in particular that there may be a role for regulators to investigate and audit key AI systems, and that this is a project the Digital Regulation Cooperation Forum is already actively considering.  The CMA also mentions the possibility of issuing guidance for certain priority sectors or functions such as comparison shopping, review and recommendation services.

Unsurprisingly the CMA also takes the opportunity to remind DCMS of the importance of the proposed Strategic Market Status regime and of implementing the broader competition and consumer reforms considered as part of the “new pro-competition regime for digital markets”. The CMA highlights this would allow it to consider “targeted and evidence-based remedies relating to AI”.

AI a competition priority

The policy debate will continue, next through the DCMS white paper and consultation later in the year. But the CMA has clearly made AI a priority. Firms with some form of market power through AI or other algorithms – even if not dominant in any primary product or services markets – should be aware of the CMA’s focus on this area and the possibility of future regulatory intervention.