마이크로칩 8월
This page was machine-translated and may differ from the original. View original

Teori Unveils Achievements of AI Code Analysis Solution 'Zint Code'

Google 우선 소스Published2026.04.29 11:47


Antropic 'Mythos' Public Vulnerability Recreated
12 Additional Vulnerabilities in the Same Codebase Identified
Theori has released the verification results for its AI-based code analysis solution, 'Xint Code.' By analyzing the same codebase as Antropic's security-specialized AI, 'Claude Mythos Preview,' it presented results demonstrating the reproduction of existing vulnerabilities and the identification of additional vulnerabilities.

Theory stated in a white paper released in mid-April 2026 that by applying the Zint code to codebases analyzed by the Mythos Preview, such as FreeBSD, OpenBSD, FFmpeg, and Firecracker, it reproduced four previously disclosed major vulnerabilities and identified an additional 12 medium-to-high risk vulnerabilities.

The subject of comparison, Claude Mythos Preview, is a cybersecurity-specialized AI model announced by Antropic on April 7, 2026. Antropic explained that it verified the potential for detecting and exploiting zero-day vulnerabilities in major operating systems and software through the model.

Through these analysis results, Theory emphasized the importance of an analysis framework that encompasses not only the performance of the specific AI model itself but also code exploration, vulnerability screening, false positive elimination, and the derivation of actionable results.

Antropic is providing the Mythos preview in a limited research environment and is simultaneously conducting a project to expand access to major technology companies and security agencies.
본 기사에 대한 정정·반론·추후보도 청구는 보도 청구 안내를, 그간 게재된 보도문은 정정·반론보도 모아보기를 참고해 주세요.
배종인 기자
배종인 기자