🇹🇼 Chinese, 5 November 2026, AI security
OpenAI Releases New Cybersecurity Model, But Is It Enough
OpenAI's new ChatGPT Cyber model has identified 948 potential vulnerabilities in Linux kernels and 81 new vulnerabilities in Firefox.
OpenAI has released a new AI model focused on cybersecurity, called ChatGPT Cyber, which is being made available to high-end users of its Daybreak Red project. The model has been used to scan Linux kernels and identify 948 potential vulnerabilities, as well as discovering 81 new vulnerabilities in Firefox.
OpenAI has also announced a $1 billion plan to support research into the security of critical infrastructure, but the industry is questioning the scale of its investment and the limitations of the project.