We, the undersigned civil society organisations, call for an immediate global moratorium on the development, deployment and sale of ‘frontier AI’ which are advanced or autonomous AI models that have the potential to cause severe or widespread harm to humans until binding safeguards, independent oversight and effective enforcement mechanisms are in place.

Developers from some of the world’s largest AI companies now accept the need to slow the reckless development of the most powerful and potentially dangerous AI systems. This is a significant admission from the companies leading the race: AI capabilities are advancing faster than the safeguards needed to control them.

Voluntary commitments are not enough. After each warning, companies return to accelerated development while governments continue to discuss future regulation. A handful of private companies cannot be allowed to determine what level of danger the public should accept.

The harms are already here

Experts warn that advanced AI could pose an existential threat to humanity. Meanwhile, AI is already enabling mass fraud, accelerating surveillance and weapons systems, amplifying misinformation, eliminating entry-level jobs and concentrating power and wealth.

We work across digital rights, labour rights, corporate accountability, gender justice, racial justice, education, environmental justice and civil liberties. We have consistently documented these harms, sounded the alarm over the reckless pace of AI development and called for binding regulation. The tech giants have fought to block effective and binding regulations which remain far from the standards applied to other sectors capable of causing severe human rights and environmental harm, including pharmaceuticals, aviation and ‘controversial weapons’.

Our urgent call

As public consensus grows on the need for urgent action, we demand that:

1. Governments impose an immediate global moratorium on the development, financing and sale of frontier AI models and begin negotiating a binding international agreement.
2. AI companies immediately cease the development, deployment and sale of frontier AI models and cooperate fully with independent regulators and assessors.
3. Investors pause financing for frontier AI, disclose their exposure and use their leverage to require portfolio companies to stop development, deployment and sale.

The moratorium must remain in place until five conditions are met:

1. Independent oversight and certification: Advanced and autonomous models must undergo independent assessment against internationally agreed safety and human rights standards before release. Regulators must have access to the systems and information necessary to assess them. Certification must include effective containment and shutdown mechanisms.
2. Transparency on safety: Companies must disclose their spending on AI development and safety, alongside serious incidents, identified risks, failed tests and corrective action.
3. Mandatory and ongoing human rights and environmental due diligence: Companies must identify, prevent and address harms across the AI value chain. These duties must be backed by civil and criminal liability for companies and their executives.
4. Prohibition of dangerous capabilities: Governments must prohibit systems with capabilities that cannot be reliably controlled, including recursive self-improvement that could outrun our ability to contain it.
5. Strict controls in critical sectors: AI must not be deployed in health, finance, public services, education or other high-risk sectors as well as migration management, weapons and surveillance systems without extensive impact assessment and independent oversight.

Action cannot wait

The world cannot wait for the United States and China to determine the future of AI. Governments across regions must work together to build an inclusive international agreement in which workers, human rights defenders and affected communities worldwide have genuine decision-making power.

In parallel, national and regional regulation must proceed rapidly. Governments have an immediate and a primary obligation to protect their populations and cannot delay action while international negotiations continue. Development of frontier AI must stop until binding safeguards are in place.

Signatories

Organisations:

1. 7amleh – The Arab Center for the Advancement of Social Media
2. Action Group on Erosion, Technology and Concentration (ETC Group)
3. Africa Freedom of Information Centre (AFIC)
4. Asociación Ciudad Viva, Paraguay
5. Bangladesh NGOs Network for Radio and Communication (BNNRC)
6. BlueLink Foundation
7. Business and Human Rights Centre
8. Cairo Institute for Human Rights Studies
9. Campaign on Digital Ethics (CODE)
10. CCAOI
11. Centro Nacional de Comunicación Social (Cencos)
12. CITAD
13. Collaboration on International ICT Policy for East and Southern Africa (CIPESA)
14. Common Room Networks Foundation (Common Room)
15. Digital Empowerment Foundation
16. Digital Justice Network
17. Digital Rights Nepal
18. Empower
19. EngageMedia
20. Fantsuam
21. Fundación porCausa
22. Foundation for Media Alternatives
23. Globe International Center, Mongolia
24. GreenNet
25. Institute for Digital Rights
26. International Service for Human Rights
27. IT for Change
28. JCA-NET.
29. Jokkolabs Banjul
30. LaLibre Tecnologías Comunitarias
31. Latin American Campaign for the Right to Education (CLADE)
32. Media Matters for Democracy
33. Media Rights Agenda (MRA)
34. Migrant Workers’ Action
35. Mizzima Media
36. Myanmar Internet Project
37. Nexus Research Co-operative, Ireland
38. ONG Amaranta - Chile
39. Open MIC (Open Media and Information Companies Initiative)
40. Palestinian Center for Development and Media Freedoms (MADA)
41. Pangea.org
42. Paradigm Initiative (PIN)
43. PEN International
44. Pollicy
45. Public Citizen
46. Software Freedom Law Center India (SFLC.in)
47. SMEX
48. SOMO
49. Taller de Comunicacion Mujer
50. TEDIC – Paraguay
51. The Media Awareness and Justice Initiative
52. The Oversight Lab
53. Digital Action
54. Gambia Press Union (GPU)
55. Cartoonists Rights
56. Media Institute of Southern Africa (MISA)
57. Freedom Forum
58. Visualizing Impact
59. VOICE
60. Freedom of Expression Institute (FXI)
61. ASPBAE - Asia South Pacific Association for Basic and Adult Education

Individuals:

1. Richard Hill
2. Andrew Garton
3. Michel Lambert
4. Leandro Navarro
5. Soledad Luca de Tena
6. Tarek Al Ghorani