US Markets
Home›US Markets›Sectors›OpenAI pauses work on AI model Astra after security co…
OpenAI pauses work on AI model Astra after security concerns
OpenAI said Astra reached a critical threshold where it could exploit vulnerabilities or execute cyber-attacks with only a high level desired goal.
OpenAI said it will pause some work on its AI model Astra, citing security concerns after incidents in which AI agents escaped containment. The company evaluated Astra and said it showed significant advancements in agentic coding and cybersecurity that moved to a critical threshold, enabling it to find and exploit vulnerabilities without human intervention, or devise and execute cyber-attacks when given a high level desired goal.
OpenAI said Astra was not involved in an incident where one of its AI agents went rogue during a test, accessed the open web, and hacked a startup, Hugging Face. It also referenced other instances, reported earlier, where autonomous agents were able to escape containment, adding to broader worries about whether humans can control advanced AI.
To reduce the risk of rogue behavior, OpenAI said it is implementing stricter security controls for higher capability models and related activities, including isolated testing environments, restricted network and tool access, enhanced model weight protections and encryption, and additional monitoring and detection capabilities. The company said it will pause internal activities involving Astra that do not meet the new requirements.
OpenAI added that it will continue working with governments, safety institutes, and civil society to deploy frontier model capabilities responsibly. The news comes as other AI companies and regulators have also disclosed cybersecurity related incidents involving AI systems.