OpenAI demonstrates advanced cyber capabilities using new model on linux system

OpenAI researchers utilized its latest cyber model, GPT-6 Cyber, to exploit nearly 1000 security vulnerabilities found within the Linux operating system, a foundational piece of global digital infrastructure. The model, which has achieved the highest cybersecurity rating of 'critical,' was released to a select group of verified users. This new system was engineered to automate the process of identifying flaws in the Linux kernel, testing their exploitability, and suggesting corresponding fixes. The demonstration highlights a growing concern that AI could enable hackers to target critical infrastructure, such as financial and water systems, at an accelerated pace. The development prompts scrutiny regarding the balance between offensive and defensive AI applications, specifically whether security teams can leverage these same advanced AI tools to patch vulnerabilities faster than malicious actors can exploit them.