ControlAI

ControlAI is an international AI-safety advocacy non-profit headquartered in the United Kingdom, calling for binding regulation on the most powerful AI systems and an outright prohibition on the development of artificial superintelligence. The organization frames the threat from uncontrolled frontier AI as comparable to nuclear risk — its public materials position extinction-level harm from superintelligent systems "on par with nuclear war" — and works almost entirely through institutional channels: parliamentary briefings, written evidence to legislative committees, draft legislation, and public-engagement campaigns that translate technical AI-safety arguments into mass-political pressure on national governments.

The organization was founded in 2023 as a campaign-arm offshoot of Conjecture, the London-based AI alignment startup co-founded in 2022 by Connor Leahy, Sid Black, and Gabriel Alfour. ControlAI's CEO and founder is Andrea Miotti, who previously worked on AI policy at Conjecture before standing up ControlAI as a dedicated advocacy vehicle; Connor Leahy serves as US Director, and Gabriel Alfour as an Advisor, anchoring the organizational lineage from the Conjecture team. Its earliest public-facing action was hiring a blimp to fly over the inaugural AI Safety Summit at Bletchley Park on 1–2 November 2023, with messaging calling on summit attendees to act on extinction risk.

Structure

ControlAI is structured as two affiliated legal entities: a UK company limited by guarantee (the originating not-for-profit) and a US 501(c)(4) social welfare organization, with consulting program officers in Canada (Samuel Buteau) and Germany (Benjamin Balde) extending its lawmaker-briefing work across four countries. The senior team layer that has emerged publicly includes Mathias Bonde (Head of Advocacy), Leticia García Martínez (UK Parliamentary Engagement Lead), Grace Gonzales (Media Engagement Lead), Sophie Toura (Operations & Outreach Manager), Max Salmon (Campaigns Strategist), Alex Amadori (Senior Policy Analyst), and a policy-research bench that includes Tolga Bilge and Adam Shimi. Advisors include the Rt Hon. the Lord Browne of Ladyton — former UK Defence Secretary, vice-chair of the Nuclear Threat Initiative, and co-founder of the European Leadership Network — and Gabriel Alfour. The organization has no chapter or membership structure; its operating model is closer to a small, full-time policy-advocacy team than to a federated grassroots organization, and it explicitly does not organize street protests.

Programs and recent work

ControlAI's flagship policy product is A Narrow Path, a long-form plan for AI governance authored by Miotti, Tolga Bilge, Dave Kasten, and James Newport and published on 19 March 2026. The plan proposes a three-phase trajectory — a twenty-year moratorium on the development of artificial superintelligence (Phase 0: Safety), an international oversight architecture to hold the moratorium against geopolitical pressure (Phase 1: Stability), and the eventual development of "safe-by-design" transformative AI under human control (Phase 2: Flourishing) — built around a compute-threshold licensing regime triggered at training runs above 10^25 FLOP. The plan has been endorsed by figures including MIT's Max Tegmark, Toby Ord, and Gary Marcus.

Its single most visible UK-specific initiative is a parliamentary-briefing campaign launched in September 2024 under UK Parliamentary Engagement Lead Leticia García Martínez. The team has systematically briefed more than 250 lawmakers across the US, UK, Canada, and Germany, and asked UK parliamentarians to sign a one-paragraph campaign statement calling for binding regulation on the most powerful AI systems. By early 2026 the statement had passed 100 cross-party UK signatories — described by the organization as the first time such a coalition of parliamentarians had publicly acknowledged the extinction threat from advanced AI. Named signatories span Labour (Anneliese Dodds, John McDonnell), Conservative (Sir John Whittingdale, Viscount Camrose — former AI Minister), Liberal Democrat (Wera Hobhouse), Green (Carla Denyer), Plaid Cymru (Ben Lake), SNP (Kirsty Blackman), SDLP (Claire Hanna), and Alliance (Sorcha Eastwood), alongside peers including Lord Browne of Ladyton and Baroness Kidron OBE and faith leaders such as the Lord Bishop of Oxford. The team has also prepared a draft bill with parliamentary lawyers and submitted it to No. 10, inviting MPs to champion the framework. Beyond the campaign statement, ControlAI has submitted written evidence to UK parliamentary inquiries on AI regulation.

Position in the movement

ControlAI sits at the institutional-advocacy pole of the AI-safety wing of the make-AI-good movement. The comparative landscape of frontier-AI organizing groups distinguishes three working theories of change: civil-disobedience tactics at lab sites (Stop AI), street-protest-and-public-letter mass-political pressure (PauseAI), and ControlAI's Westminster-inside-game — explicitly choosing institutional briefing, cross-party coalition-building, and draft legislation over visible public action. The organization's posture is also notable for routinely drawing public endorsements from credentialed figures outside the typical AI-safety crowd; the front page currently features Yuval Noah Harari and Stuart Russell among others. Its distinguishing contribution to the broader movement, on its own framing, is the translation of technical extinction-risk arguments into a parliamentary-legible policy package — compute thresholds, licensing regimes, and a phased moratorium — that elected officials can sign onto without committing to a full technical worldview.