AVERI conducts research and pilot audits to develop standards, policies, and open-source tools for independently verifying frontier AI developers’ safety and security claims.
Products & media
BenchRisk
Open-source tool and methodology that help evaluators identify benchmark failure modes, compare reliability risks, and document mitigations.
Pilot Audits
Voluntary pilots with frontier AI developers test how independent auditors can verify adherence to safety and security frameworks.
Cosigns
Patricia PaskovDirector of Standards at AVERI focused on frontier AI auditing, evaluations, and policy
“Would be remiss not to mention how much I love this team and the work that we do”
