Full Disclosure: A project is publishing full analyses of AI-discovered 0-days (seclists.org)

A new project called "0day Rubbish" has emerged, dedicated to publishing comprehensive analyses of vulnerabilities, specifically 0-days, discovered through an AI-driven research process involving a multi-LLM ensemble including Claude and OpenAI. This groundbreaking initiative has released its first batch of ten previously undisclosed vulnerabilities, with severity ratings ranging from CVSS 7.3 to 9.8, affecting eight different enterprise products. Each advisory includes a detailed technical analysis, a working exploit script, and clear reproduction steps, aiming to empower both vendors and security professionals to address these critical threats proactively. This project is significant for the AI/ML community as it exemplifies the integration of artificial intelligence into cybersecurity research, demonstrating how AI can identify vulnerabilities at scale. The disclosure model is designed to promote a transparent security ecosystem, allowing developers to verify and patch vulnerabilities while enabling cybersecurity defenders to enhance their detection and mitigation strategies. The use of automated AI in identifying such high-risk vulnerabilities raises discussions around ethical disclosure practices but also showcases the potential of AI as a tool for enhancing overall software security.