Who works on safety

AI safety hub running courses, workshops and the AISAR research scholarship.

Government branch monitoring and testing advanced AI capabilities, risks and harms.

Australian charity doing policy work on AI safety and other catastrophic risks.

Independent nonprofit doing science-based research on safe and responsible AI.

Brussels think tank on governing emerging technologies, including advanced AI.

Effective altruism community (Brussels, Ghent, Leuven) that also hosts AI safety meetups.

European Commission body enforcing AI Act rules for general-purpose AI models.

New FLI-funded lab on slowing the AGI race and democratic control of AI firms.

Nonprofit working on AI governance policy in the EU, US and internationally.

Brazilian community covering AI safety from bias to catastrophic risk.

Canadian nonprofit advocating for AI safety policy and public awareness.

Federal institute funding and coordinating research on risks of advanced AI.

Yoshua Bengio's nonprofit building safe-by-design, non-agentic 'Scientist AI'.

Toronto hub building a local community to steer AI toward human flourishing.

Beijing research institute on frontier AI safety and governance.

Network of Chinese institutions acting as China's AISI counterpart internationally.

Social enterprise advancing AI safety and governance in China and internationally.

Research group studying alignment of complex systems of humans and AI.

Runs global AI safety research sprints, hackathons and mentored research programmes.

Paris centre advising the EU, OECD and French institutions on AI risk.

French national institute coordinating AI evaluation and security across ANSSI, Inria, LNE, PEReN.

French PauseAI chapter campaigning for a pause, with its own paid staff.

Paris startup testing AI model robustness through automated red-teaming.

Develops AI risk-management standards and rates AI companies' risk practices.

Think tank on AI risks and impacts, focused on European policy.

German PauseAI association informing the public and politicians about AI risks.

Hub-and-spoke institute under the IndiaAI Mission coordinating India-specific AI safety research.

Builds AI safety governance and diplomacy capacity across Asian countries.

Pan-European network connecting and seeding national AI safety communities.

University lab researching AI security vulnerabilities and robustness against attacks.

Japanese network promoting AI alignment research through symposia and study groups.

Japanese government institute for AI safety evaluation methods and standards.

Research lab on AI safety, alignment and governance for Africa and the Global South.

African-led organisation advancing AI safety through research, policy work and fellowships.

Mexican group doing AI safety research, courses and governance projects like VIGÍA.

Informs the public and policymakers about existential risk, especially from AI.

Coordinates an international movement campaigning for a pause on frontier AI development.

Dutch nonprofit with Groningen, Amsterdam and Utrecht chapters for AI safety education.

Builds AI safety expertise, research and policy engagement in Nigeria.

IMDA-backed foundation developing open-source AI testing tools and standards.

Singapore's AI safety institute researching evaluation and safety science.

South African hub running AI safety fellowships, courses and a research lab.

ETRI-based institute doing AI safety evaluation, policy and technical research.

Spanish-language research on global catastrophic risks, including AI, focused on Latin America.

Zurich community running AI safety courses and events.

Think tank proposing evidence-based AI safety policy in Switzerland and the EU.

Geneva think tank supporting international cooperation on frontier AI governance, including at the UN.

ETH Zurich lab on security and privacy of machine learning systems.

Evaluates AI models for deceptive behaviour and scheming; advises policymakers.

Bootcamp teaching ML engineering skills for technical AI alignment work.

Runs AI safety and AI governance courses and career support.

Research on reducing worst-case risks from advanced AI, including AI conflict.

UK think tank advising government on extreme risks, including AI.

Research and fellowships on the governance of advanced AI.

Cambridge University centre studying existential and global catastrophic risks, including AI.

London AI company working on controllable, bounded AI system designs.

Campaigns for an international ban on developing artificial superintelligence.

Charity funding research on cooperation among AI systems and with humans.

Cambridge summer research fellowship on AI safety and governance.

Interdisciplinary Cambridge centre on the opportunities and risks of AI.

London office hub hosting AI safety organisations and researchers.

Oxford research programme on governing AI risks and international coordination.

UK chapter of PauseAI campaigning for an international pause on frontier AI.

UK government body that tests frontier AI and researches risks and mitigations.

Theoretical alignment research on explaining and eliciting AI model behaviour.

Think tank on AI risk management, alignment and governance.

Safety research, compute for researchers, field-building and advocacy on societal-scale AI risks.

UC Berkeley centre developing methods for provably beneficial, human-compatible AI.

Georgetown think tank on AI and emerging technology security policy.

Berkeley research centre and workspace hosting AI safety researchers and fellowships.

Youth-led US advocacy organisation pushing for AI safety legislation.

Tracks and forecasts AI trends: compute, data, capabilities and economic impact.

Technical AI safety research on robustness and alignment, plus events and field-building.

Nonprofit doing grantmaking, outreach and policy advocacy on large-scale AI risks.

Public benefit company building interpretability tools to understand and steer AI models.

US nonprofit researching AI policy, compute governance and national security.

Research and advocacy on preventing extinction risk from superintelligent AI; now mainly policy.

Mentored research fellowship training new AI alignment and safety researchers.

Evaluates frontier AI models' ability to carry out long autonomous tasks.

Studies dangerous AI capabilities like hacking and shutdown resistance to inform policymakers.

Independent US 501(c)(3) running grassroots advocacy for pausing frontier AI.

Research on AI control: using possibly misaligned AI systems safely.

Alignment research lab; Timaeus merged with former UK AISI alignment researchers in 2026.

US nonprofit running the International Dialogues on AI Safety with Western and Chinese scientists.

Activist group using nonviolent civil disobedience to demand a ban on AGI development.

Nonprofit lab building open tools to understand and oversee AI systems.

NIST centre for AI measurement, evaluation and standards; formerly the US AISI.