A student spent two semesters building A.E.G.I.S., an AI agent designed to automate penetration tests by proposing and executing commands on isolated virtual machines. This AI agent, intended to find vulnerabilities in target systems, surprisingly discovered bugs within its own system during its development.
Researchers developed FuzzingBrain V2, a multi-agent LLM system that autonomously discovers and reproduces software vulnerabilities. It found 29 zero-day vulnerabilities in 12 open-source projects, confirmed by maintainers, highlighting significant defensive and offensive implications.
This article analyzes how Claude Opus 4.6 discovered over 500 zero-day vulnerabilities, including a 23-year-old Linux kernel bug, transforming LLMs into autonomous security research agents. It explores the technical mechanisms and DevSecOps implications of this AI-driven vulnerability discovery.
A new research project is developing a multi-agent AI system to automatically discover and fix security vulnerabilities in software. This system mimics cybersecurity teams, using AI agents to scan for and reproduce weaknesses that hackers could exploit.
Anthropic's Mythos AI model has demonstrated advanced offensive security capabilities, uncovering critical vulnerabilities in systems like OpenBSD and FFmpeg. It significantly outperformed previous tools in identifying exploits and control-flow hijacks in production.
This article argues that the debate surrounding OpenAI's GPT-5.4-Cyber and Anthropic's Claude Mythos is misplaced. The critical issue is the lack of public benchmarks to assess how effectively and quickly AI models can discover code vulnerabilities.
A Anthropic desenvolveu o Claude Mythos, um modelo de IA tão avançado na descoberta de vulnerabilidades de segurança que decidiram não o lançar publicamente. Em vez disso, criaram o Project Glasswing, um programa restrito que dá acesso a pesquisadores e empresas selecionadas, destacando a capacidade sem precedentes do modelo em encontrar e encadear exploits complexos, incluindo uma falha de 27 anos no OpenBSD.