Artificial Intelligence | UK Regulatory Outlook June 2025

UK updates 

ICO launches AI and biometrics strategy  

The Information Commissioner's Office (ICO) has launched its AI and biometrics strategy. The focus is on AI uses which it sees as higher risk: 

• Foundation model development.
• Face recognition systems – use by the police.
• Automated decision-making (ADM) – especially in the recruitment context and in public services.
• Impact of emerging areas of AI – such as agentic AI and use of personal data to infer emotions or behaviour. 

Key cross-cutting issues are: 

• Transparency and explainability: being transparent with individuals about how an AI system is using their personal data.
• Fairness, bias and discrimination: ensuring that AI use of personal data is fair.
• Exercise of rights and access to redress.
• Putting in place governance and technical measures to protect individuals from harm caused by AI uses of their data. 

Businesses now can look out for: 

• A consultation on draft updated guidance on ADM and profiling to reflect the changes set out in the Data (Use and Access) Bill (by autumn 2025).
• Publication of a statutory Code of Practice on AI and ADM.
• Guidance on police use of automated face recognition.
• A report on some agentic AI issues (in particular, on accountability and redress).
• A consultation on agentic AI issues. 

Data (Use and Access) Bill passed  

Following the conclusion of the contentious ping pong process, the Data (Use and Access) Bill finally passed both Houses of Parliament and received royal assent, officially becoming the Data (Use and Access) Act 2025 on 19 June. 

After all the wrangling, the bill now includes very few provisions about AI and copyright. The government has nine months to produce an economic impact assessment and report on some of the copyright policy options regarding AI training which were set out in the government's AI/copyright consultation, including proposals on technical measures, transparency, licensing, enforcement and AI developed outside the UK. 

See more in the Data law section. 

High Court warns lawyers about misuse of AI 

Two cases highlight the duties of England and Wales qualified lawyers when using AI. The High Court addressed the cases in a recent judgment. In both, lawyers had cited cases to the court that were fictional, hallucinated by AI systems that the lawyers had used to draft documents. The judge reviewing the cases particularly emphasised the supervisory and managerial responsibilities of more senior lawyers: 

"This duty [to check the accuracy of AI-based research] rests on lawyers who use artificial intelligence to conduct research themselves or rely on the work of others who have done so. This is no different from the responsibility of a lawyer who relies on the work of a trainee solicitor or a pupil barrister for example, or on information obtained from an internet search. 

We would go further however. There are serious implications for the administration of justice and public confidence in the justice system if artificial intelligence is misused." 

The judgment is an extraordinary read throughout but never more so than in this sentence: "It is striking that one of the fake authorities that was cited to Dias J was a decision that was attributed to Dias J." The judge concluded: "A copy of this judgment will be sent to the Bar Council and the Law Society, and to the Council of the Inns of Court. We invite them to consider as a matter of urgency what further steps they should now take in the light of this judgment." 

Self-driving vehicle pilot accelerated to 2026 

The transport secretary, Heidi Alexander, has announced that the government will fast-track plans to set up pilot services of self-driving vehicles, bringing them forward to begin in spring 2026, as part of a wider Transport AI Action Plan. The small-scale pilot will enable customers in England to use an app to book "taxi-and bus-like" services without a safety driver for the first time. A wider rollout is possible when the full Automated Vehicles Act 2024 (which enables the implementation of a new legal framework for the use of automated vehicles) becomes law from the second half of 2027. 

In the meantime, the government is seeking views on what safety standards should be sought for self-driving vehicles in the UK as required by the Act. The call for evidence closes on 1 September 2025. 

Regulation of AI in advertising 

See Advertising and marketing section 

Ofcom's strategic approach to AI 

See Digital regulation section 

EU updates 

EU AI Act: Commission launches public consultation on high-risk AI systems  

The Commission has launched a consultation on the classification of high risk AI systems under the EU AI Act. 

The consultation document does not provide much in the way of insight into the Commission's views, instead asking a series of questions aimed at giving the Commission a sense of the public's understanding of the meaning of various provisions of the AI Act, and of what sorts of relevant systems are in use in practice. Occasionally, the questions are quite useful to those undertaking analysis of the Act. For example, when asking what the respondent believes counts as a safety component of a system, or as something which will endanger health and safety if it were to malfunction, the question gives examples of functionality which might be relevant. 

The questions cover the following aspects of high-risk AI systems: 

• Classification – Article 6(1)/Annex I – concept of a safety component and of each product category listed in Annex 1.
• Classification – Article 6(2)/Annex III. Questions related to:
  • AI systems in each use case under the eight areas referred to in the annex.
  • The filter mechanism of Article 6(3) AI Act allowing exemption of certain AI systems from being classified as high-risk under certain conditions.
  • Distinction between high-risk AI system and prohibited AI (Article 5).
  • Interplay of the classification with other EU legislation.
• Classification – general:
  • Intended purpose, including interplay with general-purpose AI systems.
  • Overlaps within the classification system under Annex I and III.
• Obligations:
  • Obligations of providers and of deployers.
  • Obligations of others in the value chain (Article 25).
  • Concept of substantial modification.
• Amendment of the list of high-risk use cases in Annex III and of prohibited AI practices. The consultation is open until 18 July 2025. The responses will feed into the guidelines on high-risk AI that have to be issued by 2 February 2026. 

Commission seeks views on the use of data for AI development 

See Data law section