We track 117 organisations across 16 countries working on AI safety in its various forms: from technical alignment research to government policy, from evaluations to field-building. This directory complements our Humanoid Robotics Company Directory.
117 / 117 shown
Technical
Active
Scope
Building 'safe superintelligence' as sole product/mission.
Programs
Safe superintelligence developmentScalable alignment researchSafety-by-design AI systems
Unknown
Active
Scope
Included only because user requested; safety mission not confirmed from strong primary sources in this batch.
Programs
Healthcare AI developmentAgentic AI systems
Technical
Active
Scope
Technical research on alignment/control of advanced autonomous AI systems.
Programs
Agent foundations researchDecision theoryAlignment theoryNontrivial alignment
Mixed
Active
Scope
Reducing societal-scale risks from AI via research, field-building, and advocacy.
Programs
AI safety research grantsStatement on AI RiskField-building programsCompute cluster for safety research
Technical
Active
Scope
Technical alignment/interpretability and related research.
Programs
Eliciting latent knowledgeAlignment theory researchModel evaluations
Governance
Active
Scope
AI governance research for risk mitigation and policy design.
Programs
AI governance researchPolicy fellowshipsCompute governanceInternational AI governance
Evals
Active
Scope
Understanding capabilities/impacts of advanced AI and testing risk mitigations.
Programs
Frontier model evaluationsAI safety researchPre-deployment testingInternational safety cooperation
Standards
Active
Scope
Risk mitigation guidance and safety mechanisms for advanced AI models/systems (as stated by NIST).
Programs
AI safety guidelinesRisk management frameworkPre-deployment model testingAI safety standards development
Standards
Active
Scope
Testing, evaluation, and collaborative research to harness and secure commercial AI systems.
Programs
AI standards developmentCommercial AI testingSafety evaluation frameworksIndustry collaboration
Training
Active
Scope
Research training program in model safety: control, interpretability, oversight, evals/red teaming, robustness.
Programs
Alignment research scholars programMentorship cohortsInterpretability trainingRed teaming curriculum
Training
Active
Scope
Student-led research group reducing risk from advanced AI.
Programs
Student alignment researchAI safety reading groupsTechnical workshops
Field-building
Active
Scope
Included as a meta-resource; not an AI safety org doing safety work itself.
Programs
Interactive safety org mapOrganization directory
Field-building
Active
Scope
Meta-map; not itself doing AI safety work.
Programs
AI safety landscape mappingOrganization categorization
Governance
Active
Scope
Publishes a report cataloguing AI Safety Institutes worldwide; included as governance/meta-source org.
Programs
Responsible tech pipelineAI safety institute landscape mappingCommunity building
Evals
Active
Scope
Publishes red-teaming methodology guidance on AI safety (documented).
Programs
AI safety evaluationsInternational safety cooperationJapan AI safety standards
Governance
Active
Scope
Evidence-based AI policy informed by scientific understanding of AI risks and mitigations.
Programs
AI safety policy evidence baseResearch synthesisPublic education on AI risk
Mixed
Active
Scope
International scientific synthesis of capabilities/risks of general-purpose AI systems.
Programs
Expert synthesis on AI safetyGlobal risk assessmentInternational consensus building
Anthropic
T2
Technical
Active
Programs
Constitutional AIResponsible scaling policyInterpretability researchModel evaluations
AIGS Canada
T2
Governance
Active
Programs
Canadian AI governanceSafety policy researchRegulatory advocacy
Aligned AI
T2
Technical
Active
Programs
Value alignment technologySafe AI deployment toolsAlignment consulting
ALTER
T2
Mixed
Active
Programs
AI safety research in IsraelTechnical alignmentInternational collaboration
Astera
T2
Technical
Active
Programs
Scientific research incubationAI safety-adjacent fundingPublic benefit technology
Training
Active
Programs
AI policy researchSafety communicationPolicy advocacy
Training
Active
Programs
Global AI safety coordinationInternational safety community building
Mixed
Active
Programs
AI safety grants programOpen letters on AI riskEU AI Act advocacyExistential risk policy
Technical
Active
Programs
Value alignment researchCooperative inverse reinforcement learningHuman-compatible AI theory
Governance
Active
Programs
Public awareness of x-riskMedia engagement on AI riskPolicy advocacy
Training
Active
Programs
Career advising for AI safetyMental health supportCommunity resources
Governance
Active
Programs
Responsible AI practicesABOUT ML frameworkSafety-critical AI workstreamAI incident database
Governance
Active
Programs
OECD AI PrinciplesNational AI policy trackerAI governance best practices
Mixed
Active
Scope
Threat assessment/mitigation for AI systems; applied alignment/control; evals.
Programs
Adversarial training for safetyAlignment faking researchInterpretability researchControl evaluations
Evals
Active
Scope
Independent evaluation of frontier models for catastrophic-risk-relevant capabilities.
Programs
Autonomous capability evaluationsFrontier model threat assessmentsTask-based eval frameworks
Mixed
Active
Scope
Reducing risks from dangerous capabilities in advanced AI systems; evaluations for scheming/deception; governance guidance.
Programs
Scheming evaluationsDeceptive alignment detectionIn-context scheming benchmarks
Mixed
Active
Scope
AI safety research & education nonprofit focused on safe and beneficial frontier AI.
Programs
Adversarial robustness researchAI safety via debateRed teamingAlignment research incubation
Conjecture
T1
Technical
Active
Scope
Alignment research startup; building controllable, safe development of advanced AI.
Programs
Cognitive emulation theoryInterpretability researchCoEm alignment approach
Evals
Active
Scope
Government institute supporting safe and responsible AI development/deployment in Canada.
Programs
AI safety standards for CanadaFrontier model evaluationsSafety research grants
Governance
Active
Scope
Catalyzing Canada’s leadership in AI governance and safety.
Programs
Canadian AI governance policySafety research coordinationRegulatory frameworks
Training
Active
Scope
Online, part-time AI safety research program organizing project teams.
Programs
Research bootcampsAlignment project mentorshipField-building retreats
Governance
Active
Scope
Governance research and talent development for managing risks/opportunities from advanced AI.
Programs
AI governance researchPolicy fellowshipsCompute governanceInternational AI governance
Training
Active
Scope
Runs free courses on AI safety and governance; builds community for contributors.
Programs
AI safety fundamentals courseAI governance courseScalable safety education
Field-building
Active
Scope
Resource hub supporting AI existential safety ecosystem.
Programs
AI safety resource hubOrganization directoryReading groups coordination
SaferAI
T1
Mixed
Active
Scope
AI risk measurement, risk management ratings, standards and policy work to make AI safer.
Programs
AI risk management ratingsSafety benchmarkingResponsible scaling assessments
Technical
Active
Scope
Reorient AI research toward provably beneficial systems (mission).
Programs
Value alignment researchCooperative inverse reinforcement learningHuman-compatible AI theory
Governance
Active
Scope
AI risk governance research as part of global catastrophic risks analysis.
Programs
Global catastrophic risk modelingAI risk analysisRisk assessment frameworks
Governance
Active
Scope
Policy research challenging current AI trajectory; accountability and societal risk governance.
Programs
AI accountability researchRegulatory policyAI industry analysisWorkers and AI
Evals
Active
Scope
Academic program dedicated to AI evaluation focusing on capabilities and safety.
Programs
International AI evaluation standardsCross-border model testingSafety eval harmonization
Mixed
Active
Scope
Scientific synthesis of risks and mitigations for general-purpose AI.
Programs
Global AI safety synthesis reportExpert consensus buildingInternational risk assessment
Governance
Active
Scope
Trustworthy AI principles and global policy tracking and guidance.
Programs
OECD AI PrinciplesAI policy observatoryNational AI strategies trackerAI incident monitoring
Evals
Active
Scope
Company risk management practice ratings for frontier AI labs.
Programs
AI company safety ratingsRisk management benchmarkingResponsible scaling assessments
Technical
Active
Scope
Meta-profile; not distinct from Redwood org (kept for dedupe log).
Programs
Adversarial trainingAlignment faking researchControl evaluations
Evals
Active
Scope
Model evaluation and threat research; formerly ARC Evals.
Programs
Autonomous capability evaluationsTask-based model assessmentsThreat research
Technical
Active
Scope
Multidisciplinary research program tackling AI safety issues.
Programs
AI safety research grantsAcademic safety research coordination
Standards
Active
Scope
Publishing norms to mitigate harms and risks from AI research dissemination.
Programs
Publication norms for responsible AIDual-use research guidelines
Governance
Active
Scope
Intergovernmental standard promoting trustworthy AI principles.
Programs
AI governance principlesInternational policy standards
Evals
Active
Scope
Joint work on scheming evaluations; not a standalone org.
Programs
Scheming evaluation collaborationIn-context deception detection
Governance
Active
Scope
AI policy, national security, and emerging tech governance; safety-adjacent.
Programs
AI and national security researchEmerging technology policyAI workforce analysis
Technical
Active
Scope
Trustworthy, open AI research; safety adjacent.
Programs
Trustworthy AI developmentOpen-source AI safety toolsCommunity-driven AI safety
Field-building
Active
Scope
Funding/support for safety research (ecosystem node).
Programs
AI safety research grantsScientific computing infrastructureEmerging technology support
Mixed
Active
Scope
Academic AI research umbrella; contains safety-aligned groups (e.g., CHAI).
Programs
Foundational AI researchSafety-adjacent ML researchRobustness and fairness
Standards
Active
Scope
Industry-supported nonprofit addressing significant risks to public safety and national security from frontier models.
Programs
Responsible development guidelinesSafety best practicesAI safety fundRed teaming standards
Mixed
Active
Scope
Research on existential and global catastrophic risks, including risks from artificial intelligence (technical + governance).
Programs
Existential risk researchAI safety policyExtreme technological risk analysis
Governance
Active
Scope
Interdisciplinary research on the future of intelligence and responsible AI development/governance.
Programs
Future of intelligence researchAI narratives projectKinds of intelligenceAI ethics and society
Field-building
Active
Scope
Community that helps people navigate the AI safety ecosystem and find projects.
Programs
AI safety educational gamesPublic engagement on AI risk
Field-building
Active
Scope
Fortnightly meetings discussing AI safety papers and essays (community).
Programs
Weekly reading group sessionsAI safety paper discussions
Field-building
Active
Scope
Community infrastructure mentioned as organizer for AISafety.com reading group.
Programs
Community coordinationAlignment ecosystem development
Field-building
Active
Scope
Program empowering students to use theses as a pathway to impact (career support).
Programs
Thesis topic coachingAI safety research mentorshipAcademic career guidance
Field-building
Active
Scope
Funding node for long-term survival and flourishing projects (funding).
Programs
AI safety research grantsExistential risk fundingS-process grant allocation
Field-building
Active
Scope
Directory of funders offering financial support to AI safety projects.
Programs
Funding directory for AI safetyDonor coordination
Field-building
Active
Scope
Directory to map current AI safety research teams and gaps.
Programs
Volunteer project listingsCommunity contribution matching
Field-building
Active
Scope
Meta-post documenting AISafety.com map categories and ecosystem.
Programs
AI safety field mappingResearch landscape visualization
Arb Research
T2
Field-building
Active
Scope
Publishes an impact assessment of AI Safety Camp.
Programs
AI safety benchmarkingForecasting researchAlignment evaluation tools
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI policy researchEmerging technology analysisNational security and AI
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI policy researchNational security and AIRisk assessment frameworksTechnology governance
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI governance researchTechnology policy analysisResponsible AI frameworks
Mixed
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI safety and ethics researchData science for public goodAI governance frameworks
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI accountability researchAlgorithmic auditingPublic engagement on AIRegulatory policy
Training
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI policy researchAI governance strategyEmerging technology policy analysis
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
EU AI Act implementationAI governance coordinationGPAI model oversight
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
Global AI governance recommendationsInternational AI safety normsCapacity building
Standards
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
Safety-critical AI guidelinesIndustry safety standards
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
Bio-AI risk assessmentDual-use technology governanceCross-domain risk analysis
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
Biosecurity researchAI misuse risk analysisHealth security policy
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
Nuclear risk reductionAI and WMD riskBiosecurity governance
Mixed
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
Existential risk researchAI governance theoryMacrostrategy research
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI governance researchDigital ethicsFuture of work and AI
Governance
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI and international orderTechnology and democracyDigital governance
Mixed
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI Index reportPolicy researchInterdisciplinary AI researchAI audit tools
Evals
Active
Scope
Included as part of the AI safety ecosystem; mission verification may be needed for safety-first criteria.
Programs
AI incident trackingIncident taxonomySafety learning from failures
Governance
Active
Scope
Works on preventing misuse of advanced AI and strengthening safeguards; mission verification needed.
Programs
AI security advocacyPolicy engagement on AI risk
Governance
Active
Scope
Publishes analysis/forecasts of AI trajectories; safety-adjacent.
Programs
AI governance researchFuture scenarios analysisPolicy recommendations
PauseAI
T2
Governance
Active
Scope
Advocacy group focused on slowing AI progress until safe.
Programs
AI development moratorium advocacyPublic protests and campaignsInternational chapters
Governance
Active
Scope
EU monitoring and policy support for AI.
Programs
AI landscape monitoringPolicy analysis for EUAI uptake tracking
Field-building
Active
Scope
EU community platform; not a dedicated safety org.
Programs
Stakeholder consultationAI policy input to EUCommunity engagement
Governance
Active
Scope
AI governance think tank.
Programs
AI governance researchUN and multilateral engagementResponsible AI frameworks
Governance
Active
Scope
Catastrophic risk org with AI relevance.
Programs
AI policy for UK governmentExtreme risk policyBiosecurity and AI governance
Field-building
Active
Scope
Funder; ecosystem node.
Programs
AI safety research grantsTechnical alignment fundingAI governance grantsBiosecurity and AI
Governance
Active
Scope
AI policy research and advocacy.
Programs
Public opinion polling on AIAI policy advocacyCongressional engagement
Field-building
Active
Scope
Community forum; meta node.
Programs
Technical alignment discussionResearch publication platform
LessWrong
T3
Field-building
Active
Scope
Community platform; meta node.
Programs
Rationality community platformAI safety discussion forumResearch publication
Epoch AI
T2
Governance
Active
Scope
Tracks AI progress; safety-adjacent metrics.
Programs
AI trends forecastingCompute analysisKey trends in AI publication
Technical
Active
Scope
Research institute with safety-related initiatives.
Programs
AI for humanity researchResponsible AI developmentAI safety researchTalent training
Governance
Active
Scope
Think tank work on AI governance.
Programs
Digital governance researchAI and data governanceInternational policy
Governance
Active
Scope
AI accountability and governance work.
Programs
Open Technology Institute AI workTech policy researchAI accountability
Governance
Active
Scope
International governance partnership.
Programs
Responsible AI working groupsInternational AI governanceInnovation and commercialization
Standards
Active
Scope
International AI standardization committee.
Programs
AI management system standardsAI risk management standardsAI terminology standards
Standards
Active
Scope
Standards work for A/IS.
Programs
Ethically aligned designP7000 series AI ethics standardsAutonomous systems standards
Governance
Active
Scope
AI governance and risk work.
Programs
AI governance frameworksResponsible AI toolkitGlobal technology governance
Governance
Active
Scope
Fairness/harms; safety-adjacent.
Programs
Algorithmic bias researchCoded Bias documentaryEquitable AI advocacy
Governance
Active
Scope
AI governance/harms research.
Programs
AI and automation researchMedia manipulation studiesLabor and technology
Governance
Active
Scope
Human rights risks; safety-adjacent.
Programs
AI and human rights researchSurveillance technology advocacyBan on autonomous weapons
Governance
Active
Scope
Policy and governance of AI risks.
Programs
AI governance policyPrivacy and surveillanceFree expression and AI
Evals
Active
Scope
Incident tracking; evaluation data.
Programs
AI incident trackingIncident taxonomy developmentSafety learning database
Governance
Active
Scope
Policy work including AI governance.
Programs
Internet governanceAI policy researchDigital rights
Governance
Active
Scope
Research on technology policy and AI governance.
Programs
AI governance researchInternet and societyEthics of AI
Mixed
Active
Scope
AI safety interest group page.
Programs
AI safety interest groupData science researchEthics advisory
Governance
Active
Scope
AI ethics & governance org.
Programs
AI and society researchAlgorithmic accountabilityPublic deliberation on AI
Governance
Active
Scope
EU governance office.
Programs
EU AI Act implementationAI governance coordinationGPAI oversight