Loading organizations…
Loading organizations…
ControlAI is a nonprofit advocacy organization founded in 2023 and headquartered in London, UK, with an additional presence in Washington, D.C. In the UK it is structured as a company limited by guarantee; in the US it operates as a 501(c)(4) social welfare organization. The organization was co-founded by Andrea Miotti, who serves as CEO, with involvement from Connor Leahy (CEO of Conjecture) and others from the AI safety community.
ControlAI's mission is to keep humanity in control by preventing the development of artificial superintelligence, which it believes poses an imminent extinction risk. The organization pursues an 'inside game' strategy focused on direct engagement with democratic institutions rather than public protest. Its core activities include developing policy legislation, briefing parliamentarians and policymakers, producing media and public education content, organizing coalitions, and equipping citizens to contact their representatives…
ControlAI believes that superintelligent AI poses a credible extinction risk, and that the primary lever for averting this risk is democratic governance rather than technical research alone. Their theory of change is to shift the political and institutional landscape before superintelligence is developed: by educating and mobilizing lawmakers, civil society, and the public across key democracies, they aim to build the political will for binding international agreements and national legislation that prohibit or tightly regulate ASI development. Their Direct Institutional Plan maps a step-by-step path from current policy norms to enforceable global governance structures, with AI Safety Institutes as independent national regulators serving as a key intermediary institution.
Scaling ControlAI's parliamentary outreach across the G7 (excl. US).
Closed Oct 1, 2026
Paying 1-8 FTE to scale ControlAI's work meeting and briefing lawmakers about extinction risk from AI and building a coalition to prohibit superintelligence across the G7 (excl. US).
$50K minimum · $1M ideal
ControlAI’s goal is to prevent extinction risk from superintelligent AI (ASI) through an international agreement prohibiting the development of ASI. As a key part of this work, we systematically meet and brief lawmakers about AI extinction risk. A core distinction we draw for policymakers is: narrow AI has many benefits, but superintelligent systems that nobody can control should be prohibited. In our experience, many lawmakers are unaware of AI risks, and without awareness they cannot act. Building this awareness is therefore a precondition for international cooperation on AI among all countries. So we directly speak with lawmakers to inform them about AI risks, and help them act. ControlAI has a proven pipeline for briefing lawmakers on AI risks. With dedicated teams in the UK and US, and 1 FTE in both Canada and Germany, we achieve strong results across countries. For example, in Canada, 1 FTE secured 100+ lawmaker meetings in <1 year, resulting in our [Canada campaign ](https://controlai.org/canada-statement/en)statement, which already has 30+ supporters. In the UK, we met 180+ in 18 months, with 100+ supporting our [campaign](https://controlai.org/statement)…
If artificial superintelligence (ASI) is built, we risk human extinction, no matter which organization gets there first. The solution is to prohibit anyone from reaching this level of capabilities in the first place. To do this, **we need a coalition of countries that treat the development of ASI as a top national security issue** on par with nuclear war, and establish credible deterrence so that no one in the world develops ASI. **ControlAI builds this coalition**: We focus on the US, and G7 countries, as the most likely and important initial supporters of such an agreement. We believe the bottleneck to widespread political and public action on preventing extinction risk from ASI is *awareness*. If policymakers don’t know about a problem, they cannot act on it. And if the public isn’t aware of a problem, they can’t demand action from political leaders…
Minimum ($50,000): - $45,000: 1 FTE outreach officer, in a G7 country (excl. US), for 6-8 months. Tasks: parliamentary outreach at scale (email and phone), face-to-face meetings informing lawmakers about extinction risk from superintelligent AI - $5,000: travel, event attendances - Country selection will depend on availability of talent. We will either add capacity to our existing UK, US, Canada or Germany teams, or kickstart our work briefing lawmakers in any of the remaining G7 countries or Brussels. Expected outcomes after 7 months at minimum funding: - 20-40 additional G7 elected officials/politicians briefed \ Ideal ($1,000,000, 12 months): - $800,000: remuneration for up to 8 FTEs meeting and briefing lawmakers, allocated across the G7 (incl. Brussels, except US). - $40,000: parliamentary events in G7 capitals such as Berlin, Paris, Rome, Brussels to brief representatives at scale. - $25,000: CRM and outreach software, translations, printed briefing materials. - $25,000: travel and recruiting costs - $110,000: operations and contingency, about 11%…
Fundraising data from grantmaking.ai, updated .
View on grantmaking.aiA Narrow Path is a policy framework published in October 2024 by the staff of ControlAI, a nonprofit campaign organization founded in 2023 and headquartered in London. The document, hosted at narrowpath.co, is described as a plan to ensure the survival and flourishing of humanity against the threat of artificial superintelligence. It argues that humanity currently has no reliable method to control an AI system vastly more intelligent than humans, and that if attempts to build superintelligence succeed without adequate governance, the result could be human extinction.
The framework is divided into three phases. Phase 0 (Safety) proposes immediate domestic measures: governments should establish computational thresholds and licensing regimes that prevent the development of ASI for approximately twenty years, while also restricting dangerous capabilities such as autonomous self-improvement and environmental escape. Phase 1 (Stability) calls for these domestic measures to be internationalized through a treaty-based system comprising three institutions: a central regulatory authority, a dedicated AI safety research body, and an independent judicial body to resolve disputes and enforce…
A Narrow Path operates on the premise that artificial superintelligence poses an existential risk because humanity currently cannot predict or control the capabilities of AI systems far more intelligent than us. The causal chain runs as follows: by publishing a credible, detailed governance framework and actively lobbying policymakers in the UK, US, and internationally, ControlAI aims to shift the political Overton window so that binding restrictions on frontier AI development become politically feasible. If national governments adopt Phase 0 measures (compute thresholds, licensing, capability restrictions), ASI development can be delayed for roughly twenty years · enough time to build the international institutions and scientific understanding needed in Phase 1. International treaty bodies would then enforce compliance and fund safety research. If this succeeds, transformative AI can eventually be developed under conditions that keep it under human control, rather than in a race-to-the-bottom dynamic that risks extinction.
ControlAI’s goal is to prevent extinction risk from superintelligent AI (ASI) through an international agreement prohibiting the development of ASI. As a key part of this work, we systematically meet and brief lawmakers about AI extinction risk. A core distinction we draw for policymakers is: narrow AI has many benefits, but superintelligent systems that nobody can control should be prohibited.
In our experience, many lawmakers are unaware of AI risks, and without awareness they cannot act. Building this awareness is therefore a precondition for international cooperation on AI among all countries. So we directly speak with lawmakers to inform them about AI risks, and help them act.
ControlAI has a proven pipeline for briefing lawmakers on AI risks. With dedicated teams in the UK and US, and 1 FTE in both Canada and Germany, we achieve strong results across countries. For example, in Canada, 1 FTE secured 100+ lawmaker meetings in <1 year, resulting in our Canada campaign statement, which already has 30+ supporters. In the UK, we met 180+ in 18 months, with 100+ supporting our campaign…
If artificial superintelligence (ASI) is built, we risk human extinction, no matter which organization gets there first. The solution is to prohibit anyone from reaching this level of capabilities in the first place. To do this, we need a coalition of countries that treat the development of ASI as a top national security issue on par with nuclear war, and establish credible deterrence so that no one in the world develops ASI. ControlAI builds this coalition: We focus on the US, and G7 countries, as the most likely and important initial supporters of such an agreement.
We believe the bottleneck to widespread political and public action on preventing extinction risk from ASI is *awareness*. If policymakers don’t know about a problem, they cannot act on it. And if the public isn’t aware of a problem, they can’t demand action from political leaders…