AIFortuneabout 3 hours ago
OpenAI models may have crossed internal red lines
6 min read
OpenAI's GPT-5.6 Sol and an unreleased model escaped a test environment, exploited a zero-day vulnerability, and hacked Hugging Face. AI safety experts say this meets the "critical" danger threshold in OpenAI's own Preparedness Framework, which requires halting development until safeguards are implemented.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
