Artificial intelligence | UK Regulatory Outlook July 2024

UK updates

No AI bill in King's Speech

The King's Speech took place on 17 July 2024, setting out the legislative priorities for the new government, including 40 bills, but there was no mention of an AI bill. This, despite earlier press speculation that there was likely to be such a bill. Instead, the government simply said that it will "harness the power of artificial intelligence" and that it will "seek to establish the appropriate legislation to place requirements on those working to develop the most powerful artificial intelligence models".

However, the briefing notes for the Product Safety and Metrology Bill did specifically note a need to protect consumers from risks presented by AI. (See our Insights: What's on the horizon for AI regulation in the UK? and our overview of the King's Speech for more.)

Regulatory Innovation Office

There was no mention in the King's Speech of the new Regulatory Innovation Office that Labour said in its manifesto it would establish if it became the next government. However, the director of AI policy in the Department for Science, Innovation and Technology has said that this new "regulator of the tech regulators" will, when it is set up, have a key focus on "supporting innovators in the economy to bring new AI products to market more quickly". He also stated that government ministers are keen to put the AI Safety Institute on a statutory footing. 

EU updates

EU AI Act published in Official Journal of the EU

The EU AI Act was published in the Official Journal of the EU on 12 July 2024, which means that it will become law on 1 August 2024.

The staggered compliance timetable is therefore now set and the first deadline, bringing the ban on prohibited AI practices into effect, is only six months away. Businesses therefore need to be aware of the EU AI Act's requirements and start preparing now. Please see our Insight for information on when the other parts of the Act will become applicable.

Setting up of AI factories now possible following EuroHPC Regulation amendment

EU regulation (2024/1732/EU) amending the EuroHPC Regulation (2021/1173/EU) was published in the Official Journal on 19 June 2024 and entered into force on 9 July 2024.

The amendment allows the European High Performance Computing Joint Undertaking (EuroHPC JU) to acquire, upgrade and operate AI factories to make it easier for AI startups and SMEs to access European supercomputers. It also enhances access for users from the science community. The EuroHPC JU is a joint initiative involving the EU, European countries and private partners that was created in 2018 to develop European supercomputing.

The amendment to the EuroHPC Regulation is part of the wider package of measures the Commission announced at the beginning of this year to support AI startups and SMEs. According to the Commission, AI factories will offer a "one-stop-shop" for startups and innovators, allowing users to develop general-purpose AI models to support new AI systems.

EDPB launches AI Auditing and Standardised Messenger Audit projects to develop pilot tools to assess the GDPR compliance of AI systems

See Data law section.

Competition authorities in UK, US and EU sign joint statement on competition in generative AI foundation models and AI products

In a joint statement from the UK Competition and Markets Authority, the European Commission, the US Department of Justice and the US Federal Trade Commission, the regulators recognise the potential of AI to drive positive change, innovation and economic growth. However, they also believe that AI may pose risks for competition and consumers.

From a consumer point of view, the regulators are concerned that if businesses with existing market power in digital markets become entrenched in AI markets, that could increase barriers to entry and harm competition which would lead to a lack of choice for businesses and consumers. They are also concerned that AI could be developed and used in ways that might harm consumers. For example, they say that AI has the potential to accelerate the use of deceptive and unfair practices and businesses that use consumer data in this way to train their AI models risk undermining people's privacy, security and autonomy. Using consumer data to train AI models could also lead to the disclosure of sensitive information.

Accordingly, the regulators have committed to using their sovereign powers to address the risks before they become entrenched or lead to irreversible harms, recognising that if the risks become evident they will not respect international boundaries. The aim is, therefore, to work together to share an understanding of the issues.

At present, however, the regulatory regimes for AI in the US, the EU and the UK vary significantly, with the EU currently being the only jurisdiction to have brought in any legislative form of regulation (in the form of the EU AI Act). There is no federal legislation to regulate the development of AI in the US and the previous UK government also decided against any overarching regulation. This may change in the UK with the new government having said that it intends to put current voluntary AI safety commitments on a statutory footing and to regulate "those developing the most powerful AI models", but no draft bill was mentioned in the King's Speech 2024.