OpenAI, Anthropic, and Microsoft Sign AI Safety Declaration

On November 9th, in Paris, a group of eight nations alongside leading AI firms such as OpenAI (creators of ChatGPT) and Anthropic (creators of the Claude chatbot), software titan Microsoft, and French AI firm Mistral AI signed a joint statement on AI safety. The signatories agreed to release information about their model safety assessments, share information on identified risks, open up their systems to third party auditing, and maintain accessible “off switches” for their AI models.

While the agreement lacks specific timelines or enforcement mechanisms, and details around how these safety assessments should be conducted are yet to be ironed out, this statement represents a recognition of the issue at hand rather than a functioning oversight mechanism. If major AI labs begin sharing safety information and allowing external audits, consumers and regulators will finally have a common standard against which to evaluate corporate safety promises.