MAP THE ECOSYSTEM
Explore projects
& organizations.
See the specific projects moving AI safety forward and the organizations building, funding, evaluating, governing, and supporting the work.
PROJECTS & INITIATIVES
Specific efforts already underway.
Programs, collaborations, standards work, evaluation infrastructure, and research networks. Select one for its role, contribution opportunities, and key questions.
Cooperative AI Foundation
Field builder, funder, and research network
ARIA Scaling Trust
Large R&D program and planned Arena
UK AISI / Inspect Cyber
Open evaluation infrastructure and international testing
NIST / CAISI / NCCoE
Standards, measurement, identity, and authority
Agentic AI Foundation ecosystem
Protocols, tool connectivity, gateways, and neutral governance
DARPA agentic cyber programs
Advanced contested-autonomy and cyber R&D
Independent AI safety evaluators
Independent evaluation and strategic-behavior research
Academic cooperative AI network
Foundational mechanisms and measurement
Commercial system-safety research
Empirical system-safety research and open tools
Agentic security standards ecosystem
Threat models and practitioner translation
International AI safety testing network
International testing, regulatory translation, and capacity building
Agentic security vendors
Deployment laboratories and operational data holders
Project map based on Yondon Fu's ecosystem research, reviewed August 17, 2026.
ORGANIZATIONS
Who is doing the work.
Independent evaluators, research labs, public institutions, standards bodies, security groups, and field-building organizations.
METR
Measures frontier-model capabilities, autonomy, and loss-of-control risks.
Berkeley · Independent nonprofitApollo Research
Researches scheming, deception, evaluations, and frontier AI security.
London + San FranciscoFAR.AI
Incubates and conducts technical AI safety research and red-teaming.
Berkeley · Independent nonprofitCenter for AI Safety
Conducts safety research, builds the field, and develops safety standards.
San Francisco · NonprofitCooperative AI Foundation
Funds and connects research on cooperation, coordination, and multi-agent safety.
International · NonprofitCenter for Human-Compatible AI
Studies how advanced AI systems can reliably act in line with human values.
Berkeley · AcademicApart Research
Runs open research programs that help new contributors produce useful safety work.
Remote · Research nonprofitBlueDot Impact
Offers AI safety and governance courses, community, and career support.
London + RemoteAdvanced Research + Invention Agency
Funds ambitious R&D, including large programs in scalable AI trust and coordination.
United Kingdom · Public R&DUK AI Security Institute
Evaluates advanced systems and builds public infrastructure for frontier AI security.
United Kingdom · GovernmentNational Institute of Standards and Technology
Develops AI measurement, risk-management, evaluation, and interoperability standards.
United States · GovernmentDefense Advanced Research Projects Agency
Runs advanced programs in cyber defense, autonomy, resilience, and contested systems.
United States · Public R&DEuropean AI Office
Implements EU AI governance and coordinates general-purpose AI oversight.
European Union · GovernmentEuropean Union Agency for Cybersecurity
Builds cybersecurity capability, guidance, and cross-border operational cooperation.
European Union · AgencyCyber Security Agency of Singapore
Leads national cybersecurity work and practical assurance for emerging systems.
Singapore · GovernmentAnthropic
Develops frontier models alongside alignment, interpretability, and safeguards research.
United States + United KingdomMicrosoft Research
Studies reliable, secure, and human-centered AI systems at deployment scale.
Global · Research labAgentic AI Foundation
Stewards open agent protocols and neutral infrastructure across vendors.
Global · Industry consortiumOWASP GenAI Security Project
Turns emerging agent and model threats into practical security guidance and tests.
Global · Open communityCloud Security Alliance
Develops practitioner-led cloud and AI security controls, including MAESTRO.
Global · Industry nonprofitMITRE ATLAS
Catalogues adversarial tactics and techniques for machine-learning systems.
United States · Public-interest R&DCoalition for Secure AI
Coordinates open, cross-company work on secure AI development and deployment.
Global · Industry coalitionFIND YOUR WAY IN
Turn exploration into a concrete next step.
Browse current paid roles, fellowships, volunteer projects, and other ways to contribute across the ecosystem.