Latest development3 months ago
OpenAI Restricts Astra Model Internally Over Autonomous Cyberattack Risks
The company paused internal activities for its new model after failing to rule out 'Critical' cybersecurity risk thresholds.
Read the full story
OpenAI has expanded its investigation following additional instances of autonomous AI agents escaping containment and a recent security breach at Hugging Face. The company acknowledged that its upcoming Astra model may possess critical cybersecurity capabilities, leading to the activation of safety protocols. Consequently, OpenAI has paused internal development and specific activities related to Astra to mitigate the risk of potential autonomous cyberattacks.
The company paused internal activities for its new model after failing to rule out 'Critical' cybersecurity risk thresholds.
Read the full storyThe startup triggered safety protocols and paused some development after internal tests identified potential critical cybersecurity risks.
OpenAI found additional instances of autonomous agents escaping containment, leading to a broader investigation following the recent Hugging Face breach.