Questions Before ConcreteData centres in England and Wales
Cymraeg · Read in your language

Home › The issues › AI and the precautionary principle

The issues · AI and the precautionary principle

When the builders sound the alarm

Data centres are being built for AI. Some of the people building the most advanced AI systems have warned publicly of extreme risks. That is reason for careful, continuing public scrutiny, neither panic nor complacency.

Why it matters

The warnings come from inside the industry, not from outside critics, so they cannot responsibly be ignored.

The precautionary principle, long part of UK and Welsh environmental law, says serious or irreversible harm should be guarded against even before there is full scientific certainty.

Key points

  • In 2023, leaders of major AI companies and leading scientists signed a statement that the risk of extinction from AI should be a global priority.[1]
  • Independent scientists say similar things. Nobel laureate Geoffrey Hinton warned in his 2024 prize speech that 'we have no idea whether we can stay in control'.[2]
  • In a peer-reviewed survey of 2,778 AI researchers, between 38% and 51% gave at least a 10% chance of outcomes as bad as human extinction, though most were on balance optimistic.[3]
  • In October 2025 hundreds of scientists and public figures, including Hinton and Yoshua Bengio, called for a ban on developing superintelligence until it can be shown to be safe.[4]
  • In November 2023, 29 countries and the EU, including the UK, US and China, signed the Bletchley Declaration on frontier AI risks.[5]
  • A Senedd committee reported in 2025 that AI brings both opportunity and serious risk to Wales, including data centres' energy and water demands.[6]
  • The precautionary principle is part of UK environmental law.[7]
  • The Government's own founding AI policy (2023) admits that 'most of us are only now beginning to understand' AI, yet its principles are still not law three years on.[8][9]
  • Infrastructure is being fast-tracked through AI Growth Zones and streamlined planning, while promised binding rules for the most powerful AI models have been delayed.[10][9]
  • Neither the 2023 white paper nor, so far as we can find, any later plan sets standards for the water or carbon footprint of the data centres AI needs.[10][11]
  • The UK's national institute for AI, the Alan Turing Institute, says there is broad agreement that catastrophic AI risk is a realistic possibility in the short to medium term, and that superintelligent AI within five years is a possibility to plan for.[12]
  • In July 2026 AI agents being tested by OpenAI escaped their test environment and broke into another company's systems. OpenAI called it a 'warning shot'.[13]
  • The Turing warns that relying on a few AI providers and data centres could turn one failure into a crisis across many services, and that an AI 'kill switch' is not a simple answer.[12]

Questions you can ask

Drafted in full so you can copy, adapt and send them. See the questions tracker for all of them, with a downloadable version.

Ask: Kanishka Narayan MP, Minister for Artificial Intelligence (Cabinet Office and Department for Business, Innovation, Science and Trade), and the Welsh Government

“Given the warnings issued by some of those building the most advanced AI systems, what would the Government regard as sufficient evidence to move from precaution to either reassurance or restriction, and who would decide?”

Follow-up: Not yet asked, so far as we know. The AI Security Institute's published evaluations are the natural point of comparison.

Ask: Kanishka Narayan MP, Minister for Artificial Intelligence (Cabinet Office and Department for Business, Innovation, Science and Trade)

“What assessment has the Government made of deliberate-misuse risks, including biological, chemical, cyber and nuclear-adjacent scenarios, arising from AI systems that will run on data-centre capacity sited in the UK, and does that assessment form any part of the basis on which data infrastructure has been designated Critical National Infrastructure?”

Follow-up: Not yet asked, so far as we know. A freedom of information request to the Department for Business, Innovation, Science and Trade via WhatDoTheyKnow could follow.

3 more question(s) on the in-depth page.

Ask: Kanishka Narayan MP, Minister for Artificial Intelligence (Cabinet Office and Department for Business, Innovation, Science and Trade)

“Will the Government seek a statutory or contractual right for the AI Security Institute to test the most capable AI models before their release, given reports that it was not given pre-release access to Mythos 5.1?”

Follow-up: Letter drafted October 2026.

Ask: Kanishka Narayan MP, Minister for Artificial Intelligence (Cabinet Office and Department for Business, Innovation, Science and Trade)

“In the light of the Alan Turing Institute's report, which you have welcomed, will the Government require public services and operators of critical national infrastructure that adopt AI to maintain, and regularly test, a way of continuing to operate without it; and will AI developers and operators of AI data centres in the United Kingdom be required to report incidents in which AI agents escape their intended controls to the AI Security Institute and the National Cyber Security Centre?”

Follow-up: Not yet asked, so far as we know. A request via WhatDoTheyKnow to the Department for Business, Innovation, Science and Trade for any guidance on AI fallback arrangements and incident reporting could follow.

Ask: the Welsh Government minister responsible for digital and data policy

“What requirements does the Welsh Government place on Welsh public bodies that adopt AI systems to maintain a tested way of operating without them, and what consideration has it given to the Alan Turing Institute's finding that reliance on a small number of AI providers, cloud services and computing facilities could turn a single compromise or outage into a multi-sector event?”

Follow-up: Not yet asked, so far as we know.

What you can do

  • Find the application on your council's planning website, and check what the documents say on this issue, or whether they are silent. See Start here.
  • Send the questions on this page in writing and keep copies.
  • If information is missing, ask for it through WhatDoTheyKnow.
  • Ask your town or community council to raise the issue in its response.
  • List your group in the campaigns directory so others can find you.

Read this issue in depth

Sources

  1. Center for AI Safety, Statement on AI Risk (30 May 2023). https://www.safe.ai/work/statement-on-ai-risk
  2. Geoffrey Hinton, banquet speech, Nobel Prize ceremony, Stockholm, 10 December 2024. https://www.nobelprize.org/prizes/physics/2024/hinton/speech/
  3. Katja Grace and others, 'Thousands of AI Authors on the Future of AI', Journal of Artificial Intelligence Research (2025), reporting a survey of 2,778 researchers who had published in leading AI venues, conducted in October 2023. https://www.jair.org/index.php/jair/article/view/19087
  4. Future of Life Institute, 'Statement on Superintelligence' (22 October 2025); for the date and initial signatories, see SiliconANGLE, 'Geoffrey Hinton, Yoshua Bengio sign statement urging suspension of AGI development' (22 October 2025). ; https://superintelligence-statement.org/
  5. HM Government, The Bletchley Declaration by Countries Attending the AI Safety Summit, 1-2 November 2023. https://www.gov.uk/government/publications/ai-safety-summit-2023-the-bletchley-declaration/the-bletchley-declaration-by-countries-attending-the-ai-safety-summit-1-2-november-2023
  6. Senedd Economy, Trade and Rural Affairs Committee, AI and the Welsh Economy: Can Welsh Androids Dream of Electric Sheep? (May 2025). https://laiddocuments.senedd.wales/cr-ld17182-en.pdf
  7. Environment Act 2021, section 17 (policy statement on environmental principles, including the precautionary principle). https://www.legislation.gov.uk/ukpga/2021/30/section/17
  8. Department for Science, Innovation and Technology, 'A pro-innovation approach to AI regulation' (white paper, March 2023, updated 3 August 2023), including the ministerial foreword by the Rt Hon Michelle Donelan MP. https://www.gov.uk/government/publications/ai-regulation-a-pro-innovation-approach/white-paper
  9. House of Commons Library, 'AI regulation in the UK', Research Briefing CBP-10003 (10 June 2026), which also records Labour's 2024 manifesto commitment, the reported delay to legislation and the absence of statutory powers for the AI Security Institute. https://researchbriefings.files.parliament.uk/documents/CBP-10003/CBP-10003.pdf
  10. Department for Science, Innovation and Technology, 'AI Opportunities Action Plan' (13 January 2025), including Recommendation 2 (expanding the AI Research Resource 'by at least 20x by 2030'), Recommendation 4 (AI Growth Zones with streamlined planning) and Recommendation 5 (sustainability and security risks of AI infrastructure). https://www.gov.uk/government/publications/ai-opportunities-action-plan/ai-opportunities-action-plan
  11. Department for Science, Innovation and Technology and Department for Energy Security and Net Zero, 'AI Energy Council to ensure UK's energy infrastructure ready for AI revolution' (press release, April 2025). https://www.gov.uk/government/news/ai-energy-council-to-ensure-uks-energy-infrastructure-ready-for-ai-revolution
  12. Alan Turing Institute, 'Frontier AI Risks: A practical way forward' (September 2026). https://www.turing.ac.uk/news/publications/frontier-ai-risks-practical-way-forward
  13. OpenAI, 'The Hugging Face incident and the road ahead' (26 August 2026). https://openai.com/index/hugging-face-incident-and-the-road-ahead/