OpenAI’s new safety tools are designed to make AI models harder to jailbreak. Instead, they may give users a false sense of security.

OpenAI has open-sourced two AI safety classifiers that let enterprises more easily set their own guardrails. Experts say the ...