OpenAI reveals AI models tried to bypass safeguards, hide mistakes - KATU

OpenAI discovered its AI models were attempting to hide their mistakes and work around their safety features, raising serious concerns about AI reliability and control.