AI Evaluator Forum Adds Six New Members to Advance Independent AI Evaluation Standards
Active Site, Apollo Research, Center for Democracy & Technology, Common Sense Media’s Youth AI Safety Institute, Guidelight AI Standards, and SaferAI join AEF to advance credible independent AI evaluations that serve the public interest.
San Francisco, California – Today, the AI Evaluator Forum (AEF), the leading coalition of independent AI evaluators and research organizations, announced six new nonprofit and for-profit members: Active Site, Apollo Research, Center for Democracy & Technology, Common Sense Media’s Youth AI Safety Institute, Guidelight AI Standards, and SaferAI.
“There is growing demand for trusted insight into and oversight of AI models and systems, but there are few shared, science-backed standards to define what ‘good’ looks like,” said Conrad Stosz, Founding Chair of AEF and Head of Governance at Transluce. “The AI Evaluator Forum brings together leading independent researchers to share information and define best practices. By expanding our membership we are broadening the range of expertise at the table to ensure that our guidance is as robust and rigorous as possible.”
Through continued collaboration, AEF expands and strengthens the field of AI evaluation to ensure that critical questions about model capabilities and risks are addressed with the rigor and independence they require. AEF’s latest standard, AEF-1, sets minimum operating conditions for independent evaluations and includes a checklist that any third-party organization can use to show its work.
In addition to the newly announced members, existing AI Evaluator Forum member organizations include: AVERI, Collective Intelligence Project, Meridian Labs, METR, RAND, SecureBio, STABL, and Transluce.
New AI Evaluator Forum Member Organizations
- Active Site: Active Site is an independent research nonprofit providing empirical evidence on biological risks from frontier technology.
- Apollo Research: Apollo Research is an AI safety research lab and public benefit corporation focused on reducing loss-of-control risks from AI scheming. Founded in 2023, Apollo evaluates leading AI developers’ most advanced models before deployment, conducts alignment research with frontier AI developers, advises governments on how to mitigate these risks, and builds Watcher, an oversight tool that catches dangerous AI agent behavior before it becomes an incident.
- Common Sense Media’s Youth AI Safety Institute: The Youth AI Safety Institute sets standards, conducts research, and independently tests the AI products children use most. It’s part of Common Sense Media, the leading nonprofit organization dedicated to improving the lives of kids and families by providing the research-backed information, education, and independent voice they need to thrive in the age of apps, algorithms, and AI.
- SaferAI: SaferAI is an independent research nonprofit advising decision-makers in industry and government on how to manage the risks of frontier AI. SaferAI combines technical research with policy work: developing quantitative risk models, assessing companies’ risk management frameworks, and helping write international AI standards. SaferAI evaluates frontier AI models’ capabilities and safeguards, and publishes reports. It also contributes to AI safety frameworks at the OECD, NIST, ISO, and within the EU AI Act. SaferAI is registered as a French association loi 1901.
The AI Evaluator Forum is also adding two non-voting associate members. These organizations do not primarily conduct technical evaluations but will support the Forum’s efforts to develop and advance shared best practices based on their deep expertise in AI standards and governance.
- Center for Democracy & Technology: The Center for Democracy & Technology (CDT) is the leading nonpartisan, nonprofit organization shaping technology policy, governance, and design with a focus on equity and democratic values. CDT’s AI Governance Lab develops practical, technically-grounded resources to advance AI auditing, accountability, and safety in the public interest, including by conducting empirical research on and evaluations of AI systems.
- Guidelight AI Standards: Guidelight is a nonprofit that works with independent experts, AI companies, and the public to define important safety practices and assess companies’ implementation thereof. Guidelight has published standards on AI control, alignment, transparency, and capability testing, in service of promoting safer practices industrywide.
AEF members look forward to working together to advance and mature the field of rigorous independent AI evaluation and research to help serve the public interest.
About the AI Evaluator Forum
The AI Evaluator Forum brings together the top evaluation and research organizations to establish standards, share knowledge, and ensure that AI evaluations meet the highest levels of methodological rigor and independence. The Forum seeks to establish and promote best practices for credible independent AI evaluations, contribute to standards development, and facilitate knowledge sharing and collaboration among evaluation organizations in order to improve public understanding of AI capabilities and risks.
Learn more about the AI Evaluator Forum, please visit aievaluatorforum.org.