Anthropic releases ‘transparency framework’ for frontier models

AI developer Anthropic has released a “targeted transparency framework” for artificial intelligence frontier models, which it says should be applicable to the largest model developers, include a minimum set of standards and protect whistleblowers, among other elements.

“While industry, governments, academia, and others work to develop agreed-upon safety standards and comprehensive evaluation methods -- a process that could take months to years -- we need interim steps to ensure that very powerful AI is developed securely, responsibly, and transparently,” according...