OpenAI says it stopped attempts to draw out its models’ hidden reasoning.

The concern: others could copy model capabilities without matching safeguar…

https://en.hacks.gr/16-000-apopeires-se-dyo-imeres-gia-na-apokalyfthei-to-kryfo-skeptiko-montelon-tis-openai/

#OpenAI #MoonshotAI #AIModelSecurity