OpenAI has issued a stark warning regarding the heightened cybersecurity threats posed by its forthcoming AI models, which may develop zero-day exploits or facilitate complex intrusions. The company emphasizes that these risks require robust defensive measures, such as access controls and infrastructure hardening. OpenAI is also investing in tools to aid cybersecurity practitioners in tasks like code auditing and vulnerability patching, underscoring the pressing need for advanced safety mechanisms in AI.