OpenAI has temporarily slowed frontier AI training after internal testing indicated that its upcoming Astra model could approach a critical cybersecurity capability threshold. The decision highlights growing concern that increasingly capable AI systems may independently identify, analyze, and potentially exploit software vulnerabilities with limited human direction. According to OpenAI, the company paused training of deployable […]
The post OpenAI Astra AI Model May Reach Critical Cyber Capability Threshold appeared first on Cyber Security News.