close
Jump to content

ControlAI

Add links
From Wikipedia, the free encyclopedia
(Redirected from Draft:ControlAI)

ControlAI
FormationOctober 2023; 2 years ago (2023-10)
FoundersAndrea Miotti
TypeNonprofit
PurposeAdvocacy to prevent existential risk from artificial superintelligence
HeadquartersLondon, UK[1][2]
Locations
    • Washington, D.C., United States
    • London, United Kingdom
CEO
Andrea Miotti
Key people
Connor Leahy
Websitecontrolai.org

ControlAI is a non-profit organisation dedicated to preventing existential risk from the development of artificial superintelligence (ASI). It was founded in 2023 by Andrea Miotti, who serves as its CEO.[3] The organisation advocates for an international prohibition on the development of artificial superintelligence (ASI), while supporting continued development of narrower AI applications.[4] The organisation's advisors include former UK Defence Secretary Des Browne, who is Vice Chair of the Nuclear Threat Initiative.[5]

ControlAI operates across the United Kingdom, the United States, Canada and Germany.

History

[edit source]

ControlAI was founded by Andrea Miotti in October 2023, in the run-up to the world's first AI Safety Summit at Bletchley Park. Miotti had previously been Head of Strategy and Governance at the AI safety company Conjecture.[5] Beginning in the second half of 2025, the organisation expanded its policy outreach in the United States, Canada and Germany.[6] In March 2026 Connor Leahy, the founder of EleutherAI and former CEO of Conjecture, was appointed US Executive Director.

Bletchley Park Summit and EU AI Act (2023)

[edit source]

Ahead of the November 2023 AI Safety Summit, ControlAI campaigned for binding regulation of advanced AI, arguing that the voluntary commitments under consideration were inadequate given the extinction-level risks publicly acknowledged by AI company leaders themselves, including Anthropic CEO Dario Amodei's estimate of a one-in-four chance of AI causing human extinction. Miotti was widely quoted on this in UK media including The Guardian, iNews, the Daily Mail, and The Sun.[7]

During the final negotiations of the EU AI Act in late 2023, ControlAI campaigned against industry lobbying to remove foundation models from the regulation's scope, using mobile billboard vans in Paris and Brussels and commissioning YouGov polling that found more than two-thirds of respondents supported foundation model regulation.[8][9][10] The final text retained dedicated rules for foundation models above a defined compute threshold.

Deepfakes campaign (2023–2024)

[edit source]

ControlAI ran one of the most visible UK civil-society campaigns against AI-generated deepfakes, building a cross-party group of more than 60 UK legislators who publicly endorsed banning deepfakes across the supply chain.[11] The campaign was named explicitly in a motion of the Scottish Parliament in February 2024,[12] and was covered by The Times[13] and the Frankfurter Allgemeine Zeitung.[14] In April 2024 the UK Government announced proposed legislation criminalising the creation of sexually explicit deepfakes.[15] ControlAI is also a member of the international Campaign to Ban Deepfakes coalition.

"A Narrow Path" (2024)

[edit source]

In October 2024, Miotti and others published A Narrow Path, a policy plan proposing a phased set of domestic and international measures to prevent the development of superintelligent AI while preserving room for beneficial applications.[16][17][18]

UK parliamentary campaign (2024–)

[edit source]

From late 2024, ControlAI ran a sustained briefing campaign with members of the UK Parliament; by early 2026 the organisation reported having briefed more than 150 cross-party parliamentarians and the UK Prime Minister's office.[6] This work happened while the UK's AI policy was being criticized for delays and amid calls for the Secretary of State for Science, Innovation and Technology, Liz Kendall, to deliver stronger oversight.[19][20] In December 2025, more than 100 of these parliamentarians had signed ControlAI's cross-party statement calling for binding regulation of the most powerful AI systems.[21][22] Subsequent parliamentary activity went further, taking up ControlAI's broader call for an international prohibition on superintelligence development. In January 2026 the House of Lords held two debates on advanced AI risk, with Lord Hunt of Kings Heath (Labour) asking the Government about an international moratorium on superintelligence and Lord Fairfax of Cameron (Conservative) addressing risks from autonomous AI systems; both peers thanked ControlAI on the record for support in preparing their speeches.[23] Labour MP Alex Sobel and Lord Hunt also published op-eds endorsing an international prohibition on superintelligence development.[24][25] ControlAI has presented a draft AI bill prohibiting superintelligence at 10 Downing Street[6] and in March 2026 advised on and backed an amendment to the UK's Cyber Security and Resilience Bill that would give the government powers to shut down datacentres hosting AI systems posing critical national security threats.[26][27]

United States, Canada and Germany (2025–)

[edit source]

ControlAI has extended its policy outreach to multiple jurisdictions. In the United States, according to ControlAI, by 2026 it had briefed more than 100 Congressional offices, including 18 senators and House members.[28] In Canada, ControlAI's Canadian representative Samuel Buteau briefed lawmakers and testified to the Standing Senate Committee on Human Rights,[29][30][31] while CEO Andrea Miotti testified to the House of Commons Standing Committee on Access to Information, Privacy and Ethics.[32] By early 2026, the organisation reported having briefed more than 200 lawmakers across the United Kingdom, United States, Canada and Germany.[6]

The organisation has monitored and warned against the international race for artificial general intelligence, particularly between the United States and China. ControlAI has claimed potential risks associated with President Trump's Genesis Mission, an executive initiative aimed at deploying AI across the U.S. scientific ecosystem. The organisation warned that the mission could fail "if it becomes a mere subsidy for private companies building superintelligent AI that threatens national and global security".[1]

Reception

[edit source]

ControlAI's work has been covered by outlets including The Guardian,[21] Time,[10] Nature,[33] The Times,[34] Euronews,[8] The Washington Post[35] and Channel 4 News.[36] Miotti has written op-eds for TIME,[37] and The Spectator,[38] and has given evidence to both the UK Parliament[39] and the Canadian House of Commons.[32]

The organisation's central claim that superintelligent AI poses an extinction-level risk is shared by researchers including Nobel laureate Geoffrey Hinton and Yoshua Bengio, as well as the CEOs of OpenAI, Anthropic, Google DeepMind and xAI, many of whom signed the 2023 Statement on AI Risk. The claim is disputed by others, including Gary Marcus, who argues that current AI systems are not on a clear path to superintelligence.[4][33]

See also

[edit source]

References

[edit source]
  1. 1 2 Gibney, Elizabeth; Witze, Alexandra; Ahart, Jenna (26 November 2025). "Trump's AI 'Genesis Mission': what are the risks and opportunities?". Nature. Retrieved 12 May 2026.
  2. ↑ Gibney, Elizabeth (1 December 2025). "China wants to lead the world on AI regulation — will the plan work?". Nature. pp. 503–505. doi:10.1038/d41586-025-03902-y. Retrieved 12 May 2026.
  3. ↑ Down, Aisha (25 October 2025). "AI models may be developing their own 'survival drive', researchers say". The Guardian. ISSN 0261-3077. Retrieved 23 May 2026.
  4. 1 2 "Will "superintelligence" be the last thing humans invent?". BBC Arabic. 15 November 2025. Retrieved 12 May 2026.
  5. 1 2 "About Us | ControlAI". ControlAI. Retrieved 22 May 2026.
  6. 1 2 3 4 "2025 Impact Report | ControlAI". ControlAI. Retrieved 23 May 2026.
  7. ↑ Coverage and commentary in this period included:
  8. 1 2 "Experts urge EU governments to regulate AI instead of companies". Euronews. Archived from the original on 9 March 2026. Retrieved 23 May 2026.
  9. ↑ "Former French Official Pushes for Looser AI Rules After Joining Startup". Bloomberg News. Archived from the original on 9 September 2025. Retrieved 23 May 2026.
  10. 1 2 Perrigo, Billy. "Exclusive: Brits Want to Ban 'Smarter Than Human' AI". Time. Archived from the original on 24 July 2025. Retrieved 23 May 2026.
  11. ↑ "Deepfakes Campaign Statement | ControlAI". ControlAI. Retrieved 23 May 2026.
  12. ↑ "ControlAI Campaign, The Scottish Parliament". www.parliament.scot. Retrieved 23 May 2026.
  13. ↑ Blakely, Rhys (1 March 2024). "How easy is making a deepfake audio? All I needed was six minutes online". The Times. Retrieved 23 May 2026.
  14. ↑ "Deepfakes: So gefährlich ist KI in den sozialen Medien". FAZ.NET (in German). 4 February 2024. Retrieved 23 May 2026.
  15. ↑ "Government cracks down on 'deepfakes' creation". GOV.UK. Retrieved 23 May 2026.
  16. ↑ "A Narrow Path". www.narrowpath.co. Retrieved 22 May 2026.
  17. ↑ Wood, David (25 October 2024). "A narrow path to a good future with AI, with Andrea Miotti". Tremr. Retrieved 12 May 2026.
  18. ↑ Barten, Otto (15 November 2024). "There Is a Solution to AI's Existential Risk Problem". Time. Retrieved 12 May 2026.
  19. ↑ Koopman, Saskia (8 December 2025). "MPs push for stricter AI rules as UK lags". City A.M. Retrieved 12 May 2026.
  20. ↑ Koopman, Saskia (5 September 2025). "Kendall faces tough test to rescue UK's broken AI policy". City AM. Retrieved 12 May 2026.
  21. 1 2 Booth, Robert (8 December 2025). "Scores of UK parliamentarians join call to regulate most powerful AI systems". The Guardian. ISSN 0261-3077. Retrieved 23 May 2026.
  22. ↑ "More than 100 UK lawmakers call for AI regulation to prevent extinction risk". MLex.
  23. ↑
  24. ↑ MP, Alex Sobel (15 December 2025). "'AI superintelligence regulation is what we owe voters of this generation and those to come'". LabourList. Retrieved 23 May 2026.
  25. ↑ Lord Hunt of Kings Heath (17 February 2026). "Why we need a moratorium on superintelligence research". www.transformernews.ai. Retrieved 23 May 2026.
  26. ↑ "Amendment NC12 to Cyber Security and Resilience (Network and Information Systems) Bill to Cyber Security and Resilience (Network and Information Systems) Bill - Parliamentary Bills - UK Parliament". bills.parliament.uk. 25 March 2026. Retrieved 23 May 2026.
  27. ↑ Titcomb, James (16 May 2026). "MPs demand AI 'kill switch' to defend against 'catastrophe'". The Daily Telegraph. Retrieved 23 May 2026.
  28. ↑ "Our Work in the US | ControlAI". ControlAI. Retrieved 23 May 2026.
  29. ↑ "Standing Senate Committee on Human Rights, 45th Parliament, 1st Session, Meeting 11". Senate of Canada. 23 February 2026. Retrieved 23 May 2026.
  30. ↑ Nowak, Peter (19 March 2026). "Anti-AI Advocates and Energy Companies Headline February Lobby Reports". Do Not Pass Go. Retrieved 12 May 2026.
  31. ↑ Djuric, Mickey; Lum, Zi-Ann (20 October 2025). "Conservatives: Caucus vs. base". Politico. Retrieved 12 May 2026.
  32. 1 2 Miotti, Andrea (2026). "House of Commons Standing Committee on Access to Information, Privacy and Ethics (ETHI), Meeting No. 24". ParlVu. Parliament of Canada. Retrieved 23 May 2026.
  33. 1 2 Gibney, Elizabeth (21 April 2026). "AI doom warnings are getting louder. Are they realistic?". Nature. 652 (8111): 848–850. doi:10.1038/d41586-026-01257-6. ISSN 1476-4687.
  34. ↑ "'The world is in peril': AI researchers quit with public warnings". The Times.
  35. ↑ Tiku, Nitasha (18 April 2026). "Inside a growing movement warning AI could turn on humanity". The Washington Post. ISSN 0190-8286. Retrieved 23 May 2026.
  36. ↑ "UK unemployment soars: is AI already taking our jobs?". Channel 4 News. 17 February 2026. Retrieved 23 May 2026 – via YouTube.
  37. ↑ Miotti, Andrea. "We Need a Global Movement to Prohibit Superintelligent AI". TIME. Archived from the original on 7 November 2025. Retrieved 23 May 2026.
  38. ↑ Miotti, Andrea (6 May 2026). "We are closer to AI extinction than we think". The Spectator. Retrieved 23 May 2026.
  39. ↑ UK Parliament. "Written evidence submitted by Andrea Miotti and Steven Adler (RAI0031)". Retrieved 23 May 2026.
[edit source]
Retrieved from "https://en.wikipedia.org/w/index.php?title=ControlAI&oldid=1365603042"