Astra can identify unknown vulnerabilities and develop exploits, OpenAI says. Training restarted on 28 August after a two-week pause.

OpenAI says Astra is considerably more capable at cyber tasks than GPT-5.6 Sol. Following the Hugging Face incident, it is being especially cautious about releasing it.

SAN FRANCISCO, Sept 1 : OpenAI has determined that one of its upcoming models is so capable it requires additional safety measures before it can be launched. The model, called…