OpenAI paused work on its Astra model after it warned the AI might be capable of autonomous cyberattacks
Preliminary evaluations found the unreleased model may have reached a "critical" cybersecurity threshold under the company's safety framework
Read the full story at Quartz →