Standards consortium · Research orgs · Est. 2025
The AI Evaluator Forum (AEF) is a standards-oriented consortium that brings together independent AI evaluation organizations to improve the credibility and public-interest value of third-party assessments of advanced AI systems. Its core work is to develop shared “independence, access, and transparency” expectations that evaluators can demonstrate when they publish results—especially by addressing conflicts of interest, the terms of access to systems, editorial autonomy, and disclosure of evaluation conditions.
AEF’s flagship output is AEF-1, a voluntary minimum-conditions standard and checklist intended to be published alongside evaluation results to help readers assess how independence and access constraints may have affected findings. In addition, AEF launched a “Transparency about Third-Party AI Evaluations” public statement initiative intended to raise expectations that evaluation publishers disclose key conditions such as methods control, editorial control, evaluator access/time, and conflicts-of-interest handling.
Publications are not yet indexed for this organization. Read them on aievaluatorforum.org ↗.
The AI Evaluator Forum says it welcomes recent statements emphasizing the importance of embedded independent experts for verification.
Launch Announcement: Leading AI Research Organizations Create Forum of Independent EvaluatorsAEF announced its creation as a network of independent evaluation organizations and released AEF-1 plus a public statement initiative aimed at increasing transparency about third-party evaluations. The launch post also names the forum’s founding member organizations and several featured voices.