UK Regulator Flags Competition Risks Of AI Foundation Models

The Competition and Markets Authority (CMA) signalled it has “real concerns” with AI Foundation Models (FMs) that are controlled by a small number of tech firms.

The CMA investigated Foundation Models (FMs) last year as the market continues to develop at a “whirlwind pace”.

That investigation last September identified risks associated with AI FMs, and proposed seven guiding principles for the foundation models that underpin AI systems.

Now the CMA in an update paper has identified an “interconnected web of over 90 partnerships and strategic investments involving the same firms: Google, Apple, Microsoft, Meta, Amazon, and Nvidia (the leading supplier of AI accelerator chips).”

Interconnected web

Speaking at a conference in Washington DC, CMA CEO Sarah Cardell described the transformative promise of FMs as a potential “paradigm shift” for societies and economies.

But she also highlighted how a range of fast-moving developments across FM markets have prompted a marked increase in concern for the UK competition regulator.

The CMA cited the “interconnected web” of AI partnerships involving the same firms: Google, Apple, Microsoft, Meta, Amazon, and Nvidia.

“When we started this work, we were curious,” said Sarah Cardell. “Now, with a deeper understanding and having watched developments very closely, we have real concerns.”

“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,” said Cardell.

“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,” said Cardell.

Competition concerns

The CMA said it recognises the huge wealth of resources, expertise and innovation capability firms such as Google, Apple, Microsoft, Meta, Amazon, and Nvidia can bring to bear, and the role they will likely have in FM markets.

The regulator also said it recognises the fact that partnerships and arrangements of this kind can play a pro-competitive role in the technology ecosystem.

However, the CMA cautioned that powerful partnerships and integrated firms should not reduce rival firms’ ability to compete, nor should they be used to insulate powerful firms from competition – particularly considering the breadth of potential use for FMs, across all sectors of the economy, such as finance, healthcare, education, defence, transport, and retail.

The CMA said it is concerned at the “winner takes all dynamics” which has led to the rise of a small number of powerful platforms, and Sarah Cardell says the CMA is “determined to apply the lessons of history” at this pivotal moment in the emergence of a new, potentially transformative technology.

Cardell in her speech highlighted three key interlinked risks to fair, open, and effective competition:

• firms controlling critical inputs for developing FMs may restrict access to shield themselves from competition;
• powerful incumbents could exploit their positions in consumer or business facing markets to distort choice in FM services and restrict competition in deployment;
• partnerships involving key players could exacerbate existing positions of market power through the value chain

The CMA’s update paper provides details on how each risk would be mitigated by its principles, as well as the actions the CMA is taking now, and considering taking in the near future, to address these concerns.

To mitigate any risk, the CMA proposed a set of underlying principles, which are:

• access: meaning ongoing ready access to key inputs
• diversity: to ensure sustained diversity of models and model types
• choice: enabling sufficient choice for businesses and consumers to decide how to use foundation models
• fair dealing: by which we mean, for example, no anti-competitive bundling, tying or self-preferencing
• transparency: in terms of consumers and businesses having the right information about the risks and limitations of models
• accountability: ensuring developer and deployer accountability for outputs

The CMA concluded that without fair, open, and effective competition and strong consumer protection, underpinned by these principles, it sees a real risk that the full potential of organisations or individuals to use AI to innovate and disrupt will not be realised, nor its benefits shared widely across society.