The AI Daily Brief · Wednesday, September 2, 2026
OpenAI announced that its forthcoming model, Astra, has met a critical cybersecurity capability threshold, indicating it can identify and exploit unknown security flaws autonomously. The model achieved a perfect score on a benchmark for exploit development and demonstrated novel exploit creation on undisclosed vulnerabilities, outperforming previous models like GPT-6 sole in efficiency and capability.
“In layman's terms, that means that the model is capable of finding and exploiting previously unknown security flaws without human guidance.”
“On this data set, Astra achieves much higher arbitrary code execution rates than GPT-5/6 sole, using far fewer output tokens.”
“During the evaluation, the model even discovered and used two zero-day vulnerabilities as part of an exploit chain.”