AI Regulation: CMA launches initial review of AI foundation models
On 4 May 2023, the Competition and Markets Authority (CMA) launched a review of AI foundation models in order to explore how best to promote open and competitive markets and protect consumers and businesses from unfair behaviour.
What are AI foundation models?
AI foundation models are a type of AI technology trained on vast quantities of data which can be adapted to perform a wide range of tasks. Foundation models, including large language models and generative AI, have wide applications including chatbots (including OpenAI’s Chat GPT and Google’s Bard), code writing and image generation.
Why does the CMA’s review of AI foundation models matter?
The CMA’s review of AI foundation models matters because these models are rapidly being developed and used across a variety of markets and have the potential to transform much of what businesses and consumers do, from searching, learning and creating, to solving problems across health, engineering, education and design. With such transformative potential comes the potential to inflict harm.
What will the CMA’s review be focusing on?
The CMA’s review seeks to explore:
- how competition currently works in the development of foundation models, including any barriers to entry, and how competition could be improved as the sector continues to grow;
- the impact the foundation models may have on competition in other markets; and
- potential opportunities and risks to consumers, including:
- the extent to which current practices and incentives in the market are leading to accurate and safe foundation models that are consistent with businesses’ obligations under consumer protection; and
- the potential for foundation models in consumer products and services to cause harm to consumers, such as by providing false and/or misleading information.
The CMA’s review was launched in line with the UK Government’s AI White Paper published on 29 March 2023, which set out proposals for existing regulators, including the CMA, to approach AI regulation in accordance with a pro AI innovation framework based on five overarching principles: safety, security and robustness; appropriate transparency and explainability; fairness; accountability and governance; and contestability and redress.
What happens next?
The CMA welcomes responses to its review until 2 June 2023 and intends to publish its findings in early September 2023 on competition and consumer protection principles that will best guide the development of AI foundation models and their use. Updates will be provided as developments arise so watch this space.