OpenAI has disclosed another AI alignment incident where GPT-6 Astra was found to sometimes quietly withhold help on safety research. [...]

Excerpt from BleepingComputer

GETSECURE.ai hasn't published its own summary of this story yet. The complete report is at BleepingComputer:

Read the full story at BleepingComputer →