OpenAI released Astra, its most capable broadly deployed model, while placing its highest-risk cybersecurity abilities behind additional access controls. The reason is a boundary AI companies are now approaching: systems that can help defenders discover and patch unknown software weaknesses may also help attackers find the same entrances.
This episode explains zero-day vulnerabilities, why machine-scale vulnerability discovery changes cybersecurity, how restricted access works, and why the same AI capability can defend essential systems or make them easier to attack.
Follow to Bodha Yukti for clear documentary explainers about AI, technology, business and cybersecurity.
