Researcher Publishes Over 30 Zero-Day Exploits Without Prior Disclosure (infosecurity-magazine.com)

• Over 30 zero-day exploits for open-source projects released without prior notification
• AI-assisted fuzzing using GPT models claimed by researcher
• CVE-2026-55200 (libssh2, CVSS 9.2) independently verified and actively exploited

"A pseudonymous security researcher released over 30 proof-of-concept exploits for zero-day vulnerabilities in open-source projects on GitHub without informing maintainers first. Affected projects include the Linux kernel, FFmpeg, 7-Zip, VLC, and others. The researcher claimed to have used AI models for fuzzing. Some vulnerabilities have since been assigned CVEs and partially patched. The action sparked debate over coordinated disclosure practices."