OpenAI's forthcoming model, Astra, has demonstrated significant cybersecurity capabilities, potentially reaching the highest risk category, where it can autonomously identify and exploit vulnerabilities, as well as launch end-to-end cyberattacks against secure targets. This assessment was made following internal testing and expert reviews, which revealed substantial advancements in agentic coding and cybersecurity. OpenAI has subsequently tightened safeguards to mitigate potential risks. The company's evaluation of Astra's capabilities has prompted a re-examination of its security protocols, highlighting the need for robust measures to prevent potential misuse. Astra's capabilities could have far-reaching implications for cybersecurity, as a system with such abilities could potentially be used to launch sophisticated attacks1. This development matters to cybersecurity practitioners, as it underscores the importance of staying ahead of emerging threats and developing effective countermeasures to address the potential risks posed by advanced AI systems.