Hugging Face applies to conduct AI safety review inside Anthropic.

Beating AI News Flash: Hugging Face CEO Clement Delangue announced the establishment of the Open Alignment Initiative, led by co-founder Thomas Wolf, and publicly applied to join Anthropic's newly proposed "Embedded Evaluator" program. If approved, Hugging Face's evaluators can be stationed on-site long-term, gaining near employee-level access to directly inspect model training and safety measures.

Delangue stated that AI alignment cannot continue to be solved behind closed doors within a handful of frontier labs and requires broader participation from external researchers. Anthropic has committed to opening such access to third parties, including workstations, door access, company computers, and tools close to internal risk teams, and allowing evaluators to independently publish their conclusions.