Công ty an ninh AI Depthfirst phát hiện các lỗ hổng Internet nghiêm trọng với chi phí thấp hơn 10 lần so với Anthropic

The Huoxing Finance report states that on May 12, AI cybersecurity startup Depthfirst announced that its proprietary AI vulnerability discovery model identified multiple critical security vulnerabilities missed by Anthropic Mythos, claiming the overall cost is only one-tenth of Mythos'. Depthfirst CEO Qasim Mithani said that by optimizing the model architecture for single-task performance, the company can achieve "doing with $1,000 what Mythos would require $10,000 to accomplish." Depthfirst also launched a new initiative called the "Open Defense Initiative," offering a total of $5 million to make its AI vulnerability detection tools available to enterprises and open-source developers for identifying code security issues. The article notes that one key vulnerability discovered by Depthfirst exists in NGINX, the world's most widely used web server; this vulnerability has existed since 2008 and could theoretically affect a large number of websites globally. F5 Networks, responsible for maintaining NGINX, is expected to release a patch this week. Additionally, Depthfirst disclosed that it found a critical vulnerability in the Linux system that could be exploited for remote code execution, which has not yet been patched. Its model also identified multiple security issues in Google Chrome and the open-source multimedia framework FFmpeg, with the Chrome-related vulnerabilities already fixed by Google. The report highlights that as AI accelerates vulnerability discovery, the cybersecurity industry is entering a phase where both offense and defense are becoming AI-driven. Previously, Anthropic revealed that its Claude model was used by a Chinese hacking group for cyberattacks, and Google also warned this week that criminal groups have begun leveraging AI to develop "zero-day" exploit tools.