OpenAI has opened access to the GPT-6 Cyber ​​AI model for verified customers to search for software vulnerabilities, Reuters reports. The developer assigned the neural network a "critical" risk level. Using the new tool, researchers have already identified 81 previously unknown vulnerabilities in the Firefox browser and developed exploits for 948 vulnerabilities in the Linux kernel.

OpenAI has opened access to the GPT-6 Cyber ​​AI model for verified customers to search for software vulnerabilities, Reuters reports. The developer assigned the neural network a "critical" risk level. Using the new tool, researchers have already identified 81 previously unknown vulnerabilities in the Firefox browser and developed exploits for 948 vulnerabilities in the Linux kernel.