As major AI firms pledge greater outside scrutiny of their systems, a coalition of eight nonprofit research groups has formed the AI Evaluator Forum to craft common rules for independent testing. The effort aims to standardize how cutting-edge models are probed for safety, security and real-world risk, positioning evaluators as a counterweight to company-led assurances. The group could shape how regulators and policymakers assess AI claims, but questions remain over authority, funding and how much access tech giants will truly grant. The rise of third-party testing highlights a broader shift toward formal accountability in the AI industry.
Related articles:
NIST AI Risk Management Framework (AI RMF 1.0)
The Bletchley Declaration on AI Safety
UK AI Safety Institute – About
European Commission: The AI Act and Europe’s approach to AI
AI Incident Database – Case Reports on Real-World AI Failures
