Anthropic has expanded its Project Glasswing initiative, significantly scaling access to its Claude Mythos Preview model as part of a broader effort to strengthen global software security.
The program, first launched in April 2026 with around 50 organizations, has now grown to include approximately 200 partners worldwide, marking a major shift in how artificial intelligence is being integrated into cybersecurity defense strategies.
The newly added organizations span more than 15 countries and include sectors that are critical to national and global stability, such as energy, healthcare, water systems, telecommunications, and hardware manufacturing.
Many of these entities maintain widely used software or infrastructure, meaning vulnerabilities in their systems could impact hundreds of millions of users.
Anthropic stated that participating organizations must meet strict security requirements before gaining access to Claude Mythos Preview. The expansion also prioritizes vendors and open-source maintainers whose codebases are deeply embedded across enterprise and government environments.
Over 10,000 Critical Vulnerabilities Identified
Since its initial rollout, Project Glasswing participants have used Claude Mythos Preview to identify more than 10,000 high- and critical-severity vulnerabilities across various codebases. The AI model is designed to analyze large-scale software environments, detect security flaws, and assist in remediation efforts.
For example, a large infrastructure provider could deploy the model to scan millions of lines of legacy code, quickly identifying memory safety issues or misconfigurations that would typically take weeks of manual auditing to uncover.
Anthropic emphasized that the role of AI in cybersecurity is rapidly evolving from detection to full lifecycle vulnerability management. In addition to identifying flaws, organizations are now using Mythos Preview to:
- Generate security patches automatically
- Perform pre-release code audits
- Simulate adversarial attacks through AI-driven penetration testing
- Assist in modernizing legacy systems using safer programming languages
To support broader adoption, Anthropic also introduced Claude Security, a commercial offering built on its frontier models such as Claude Opus 4.8. The platform enables automated code scanning and patch recommendations, targeting enterprise security teams.
Anthropic warned that “Mythos-class” AI capabilities could become widely available within 6 to 12 months, potentially without adequate safeguards. This raises concerns that threat actors may leverage similar tools to automate vulnerability discovery and exploitation at scale.
As a result, the company is working with governments, industry groups, and open-source communities to establish new security norms, disclosure frameworks, and response processes tailored for AI-driven threat environments.
Despite rapid vulnerability detection, Anthropic highlighted that patching and disclosure remain significant bottlenecks. The company is collaborating with third-party organizations to improve vulnerability triage workflows and accelerate remediation timelines, particularly in open-source ecosystems.
Anthropic plans to further expand Project Glasswing, focusing on critical infrastructure providers and cybersecurity researchers. The company is also developing a Cyber Verification Program to grant controlled access to advanced AI capabilities for defensive use cases.
As AI continues to reshape both offensive and defensive cyber operations, initiatives like Project Glasswing signal a growing effort to ensure defenders maintain a strategic advantage in an increasingly automated threat landscape.
Follow us on Google News, LinkedIn, and X to Get Instant Updates and Set GBH as a Preferred Source in Google.
