Anthropic has released an initial update on Project Glasswing, revealing that its Mythos tool has identified more than 10,000 vulnerabilities for partners, many classified as high or critical severity.
Anthropic announced progress on Project Glasswing, an initiative leveraging AI to detect security vulnerabilities at scale. The company's Mythos tool has uncovered over 10,000 bugs for participating partners, with a significant portion flagged as high or critical in severity.
The vulnerability discoveries span multiple codebases and represent real-world security issues that could pose substantial risks if left unpatched. Anthropic's update indicates the tool is successfully identifying flaws that traditional security testing methods may miss.
Project Glasswing represents Anthropic's effort to apply large language models and AI techniques to cybersecurity challenges. By automating vulnerability detection, the initiative aims to help organizations strengthen their security posture without requiring proportional increases in security team resources.
The company has not disclosed specific details about which partners are using Mythos or the nature of the codebases being scanned. However, the high and critical vulnerability classifications suggest the tool is finding genuinely impactful security issues rather than minor code quality concerns.
This development aligns with broader industry trends of applying generative AI to software security. As organizations struggle with vulnerability management at scale, AI-powered tools that can quickly parse large codebases offer potential efficiency gains.
Anthopic's disclosure comes as the company continues positioning itself as focused on AI safety and responsible AI development. Security vulnerability detection represents a practical application where AI capabilities can deliver measurable business value while remaining clearly bounded and verifiable.
The Australian Cyber Security Centre has issued an alert about coordinated exploitation of vulnerable content management systems and plugins worldwide. The campaign targets organizations using outdated or unpatched CMS software.
Artificial intelligence discovered a critical security vulnerability in Linux kernel code that human developers overlooked for over a decade. The bug could allow unauthorized root access to systems.
Researchers have demonstrated a new attack called 'Ghostcommit' that hides prompt injections in PNG files to fool AI code reviewers and agents into exposing repository secrets.