March AI policy update: Reeves bets on AI as ministers U-turn on copyright

UK updates

Reeves stakes growth on AI in Mais lecture

The chancellor, Rachel Reeves, used the Mais lecture to argue that the UK faces a choice between charting its own AI path or leaving the technology's development to other countries.

Doubling down on commitments first set out in the AI Opportunities Action Plan, the chancellor outlined plans to build sufficient compute, accelerate adoption, invest in skills and compete in areas where the UK is already strong.

However, Erin Young of the Institute of Directors cautioned that AI cannot be treated as a universal solution to productivity challenges, and that a nuanced approach grounded in sector-specific contexts, safety and public trust is essential.

Ministers ditch copyright plan after artist outcry

Following a significant outcry from major artists and singers, the government reversed course on its position on AI and copyright.

The technology secretary, Liz Kendall, said ministers had "listened" and would no longer pursue the position of allowing AI companies to use copyrighted works to train models with an opt-out option.

However, for anyone hoping for clarity, the government has said it no longer has a preferred option and is seeking views across a range of areas. For more on what insight ministers are looking for, see Ella Harvey's summary.

Government bets £40m on solving AI's flaws

The government launched a new AI research lab to unlock breakthroughs that could transform healthcare, transport, science and everyday technology.

Backed by £40m in government funding, the new lab will aim to solve problems in how AI works, such as hallucinations, unreliable memory and unpredictable reasoning. It also hopes to develop new approaches that could make future AI systems more accurate, transparent and trustworthy.

Launching the lab is a first step towards delivering the UKRI AI Strategy, which aims to help AI deliver for the UK's cutting-edge science and research efforts.

Sovereign AI Unit opens supercomputer to startups

The UK Sovereign AI Unit opened applications this month for UK startups to access the AI Research Resource (AIRR), which includes some of the country's most powerful supercomputing infrastructure.

The unit wants to hear from founders in strategically important areas, including compute efficiency, next-generation AI labs, health and life sciences AI, and AI for scientific discovery.

For more on how to apply, see Madeleine Steggall's post.

Barnsley pilots target NHS waiting lists with AI

The government announced two new pilot schemes for Barnsley, the UK's first tech town.

The Living Lab project aims to cut hospital waiting lists by helping doctors with patient summaries, and to reduce missed appointments by using AI reminders. The government also announced an £800,000 upskilling fund, run by the local council, aimed at reaching SMEs and individuals who might otherwise be left behind.

Announcing the news, the technology secretary said the pilots demonstrate the government's commitment to help local residents and businesses embrace AI and ensure NHS staff spend less time on paperwork and more time with patients.

Alexandra Wood, a senior reporter for the Yorkshire Post, has more in her article.

The £40bn open source AI prize

A new report by public policy research consultancy Public First, commissioned by Meta, argues that the UK is sitting on a £40bn opportunity if it can realise the potential of open source AI.

The report proposes pragmatic, costed policy proposals to realise this potential, including a British open source AI investment credit and a £100m sovereignty fund to help SMEs adopt AI tools and technology.

For more, see Markus Reinisch's post.

Why UK AI strategy needs a tax overhaul

AI policy covers many areas, from planning to industrial policy. However, one of the little-discussed areas that could impact the UK's ability to be a world leader is tax policy, argues Tim Sarson.

In an opinion piece for City AM, Sarson, head of tax policy at KPMG, suggests that the UK needs an IP box regime that includes innovative software applications such as AI.

Such a regime extends the concept of a patent box, which lowers a firm's tax rate to 10 per cent on certain types of income. However, as most AI models and applications are unpatented, they cannot apply for this patent box. Extending the principle to IP like software and AI may help stem the flow of large multinationals buying UK businesses and exiting their IP to the US.

Global updates

AI as the next supply chain threat

Compromise of software supply chains exploits the implicitly trusted dependencies of the software rather than the defended system itself, and it can result in significant damage before the attack is detected and resolved.

AI is built and deployed under the same dependency-heavy paradigm as software, only with higher stakes, argues Dr Melina Bekyou in a new article for the Royal United Services Institute (Rusi).

When an organisation purchases an AI-enabled product, it has limited or no insight into what lies beneath. So, any weaknesses introduced by a new AI tool can filter down into deployed systems and shape behaviour, security or reliability long after the initial rollout.

For policymakers, the implications are significant. As AI becomes more embedded in critical national infrastructure and decision-making processes, supply chain compromise becomes a plausible route for systemic disruption.

For more, see Pia Katharina Hüsch's post.

Cyber agents that plan their own attacks

The Institute for AI Policy and Strategy (IAPS) published new research this month on highly autonomous cyber-capable agents (HACCAs).

HACCAs are systems that can autonomously plan and execute sophisticated cyber operations over long time horizons without continuous human oversight. As Theo Bearman, a frontier security researcher notes, HACCAs present novel risks, including loss-of-control scenarios.

The report argues that governments have a narrow window to establish policy guardrails before autonomous cyber capabilities proliferate beyond major powers. Doing so will require coordinated action across multiple policy domains and adapting risk management frameworks for agentic systems, among other measures.

For more, see Theo's post.

New views

In conversation with Alexandru Voica

Continuing the Appraise Network's series of interviews with AI policy leaders, we spoke to Alexandru Voica, head of corporate affairs and policy at AI video platform Synthesia.

In the interview, Alex explains his journey into AI policy, shares his perspectives on current AI regulatory approaches, and outlines why prioritising opportunity is crucial for the UK and Europe.

He also discusses how the government could help AI companies by prioritising talent and skills, recommending incentives to attract workers and help graduates launch technology startups, with a particular emphasis on supporting technical founders within the ecosystem.

Tim Flagg on skills and the AI workforce

Continuing the theme of talent and skills, Tim Flagg, chief executive of UKAI, the trade body for the UK's AI sector, has been interviewed by Nikesh Gosalia for Insights Xchange.

In the conversation, Flagg discusses what governments, businesses and institutions can do in skills, education and leadership as AI continues to impact the workplace. As he notes, UKAI has been calling on the UK government to focus more on earlier and broader education, and we can do more to equip workers with the skills they need to retrain given the pace of change.

For more, see Tim's post.

Chartered PR practitioner James Boyd-Wallis is MD of tech and AI focused corporate and public affairs agency Highbury Communications and co-founder of AI policy network, Appraise.

Further reading

If AI now stands for 'additional income' what does it mean for PR?

As YouTube turns 21, here are 21 ways video is changing PR forever

How PR pros can maximise visibility for their clients in an age of AI