AI labs clash with evaluators over publication rights
Anthropic and OpenAI have sought control over what independent AI safety evaluators can publish. The companies are concerned about protecting valuable trade secrets and proprietary data during testing.
Xurve View
Insights:
Xurve View
Anthropic and OpenAI sought significant editorial control over independent safety reviews to protect trade secrets, raising tensions over how powerful AI models are scrutinized. The talks highlight friction between making dangerous technologies transparent and safeguarding intellectual property for firms eyeing trillion-dollar valuations. The outcome will shape public oversight as AI advances rapidly, according to people familiar with the matter.
