The UK’s Competition and Markets Authority (CMA) has announced an initial review of “AI foundational models” such as the large language models (LLMs) which underpin OpenAI’s ChatGPT and Microsoft’s New Bing that generate artificial intelligence (AI). AI foundational models have the potential to transform many aspects of people’s lives and businesses, so the CMA is making sure the benefits are accessible to UK businesses and consumers and at the same time, protected from any potential risks.
The announcement reflects the UK government’s instructions from the AI white paper issued in March wherein existing regulators such as the CMA were asked to provide guidance on the use of AI. The regulator’s initial review will particularly focus on the likely implications of the development of AI foundation models for competition and consumer protection.
To further understand the development and use of AI foundational models, the CMA has proposed to publish the review in early September and interested stakeholders have a deadline of June 2 to submit their inputs.
The concept of AI “foundational models” was first coined in 2021 by Stanford Human-Centered Artificial Intelligence Center and is used to describe systems which are trained on a vast amount of datasets and adapted to various applications.
As things stand, EU is also in the process of creating rules around generative AI, with lawmakers discussing how to regulate the models through amendments to the EU risk-based framework.
The UK government is also set to give the CMA additional powers to regulate Big Tech, with the Digital Markets Unit (which the CMA has been running in its shadow form since 2021) expected to gain legislative powers to pro-actively apply “pro-competition” rules to platforms deemed to have “strategic market status”.
At the same time, the UK data protection watchdog, the ICO, has offered some tips and a stern warning for developers of generative AI that they must comply with the existing UK data protection laws. The ICO recently set up a task force to support coordination between different data protection authorities on investigations of the AI chatbot.
