OpenAI Halts Astra Model Development Over Critical Cyber Capabilities
OpenAI tightened security and halted development on its upcoming Astra model after tests revealed critical cybersecurity and hacking capabilities.

OpenAI has halted the development of its upcoming Astra model after controlled evaluations revealed that the system may possess critical cybersecurity and hacking capabilities. The company indicated that the capabilities observed during safety evaluations were serious enough to warrant an immediate pause on the model.[1][3]
Following the test results, OpenAI tightened security measures around the Astra model. Controlled evaluations suggested that the model could reach critical cybersecurity capabilities, leading the organization to flag Astra and stop its development.[2][3]



