Anthropic’s advanced AI cybersecurity model, Claude Mythos Preview<\/strong>, has identified more than 10,000 high- and critical-severity software vulnerabilities<\/strong> in a single month \u2014 a discovery rate that has overwhelmed the open-source community’s capacity to respond and forced a rethink of how the software industry manages security at scale.<\/p>\n<\/p>\n Project Glasswing<\/strong> is a defensive initiative launched by Anthropic to secure critical global software infrastructure. It grants approximately 50 partner organisations<\/strong> exclusive early access to Claude Mythos Preview \u2014 a frontier model capable of autonomously identifying vulnerabilities in widely-used software before malicious actors can exploit them.<\/p>\n<\/p>\n Anthropic published its first quantified results on 22 May<\/strong>, one month after the initiative launched with twelve named partners. The programme is anchored primarily in the United States and the United Kingdom, with participants including Cloudflare<\/strong>, Mozilla<\/strong>, the UK AI Security Institute<\/strong>, and security research firm XBOW<\/strong>.<\/p>\n<\/p>\n Most participating organisations \u2014 which maintain software critical to internet and essential infrastructure \u2014 found hundreds of high-risk vulnerabilities within weeks. Several reported bug-finding rates increasing by more than a factor of 10<\/strong>.<\/p>\n<\/p>\n The UK AI Security Institute<\/strong> confirmed Mythos Preview as the first model to solve both of its cyber range simulations \u2014 multistep cyberattack scenarios \u2014 end to end.<\/p>\n<\/p>\n Among the most consequential discoveries was CVE-2026-5194<\/strong>, a critical vulnerability in the WolfSSL cryptographic library<\/strong> \u2014 a widely used embedded TLS library deployed across IoT devices, automotive systems, and industrial control environments. The flaw carried a CVSS score of 9.1<\/strong> and could allow an attacker to forge certificates and impersonate legitimate services. It has since been patched.<\/p>\n<\/p>\n What distinguishes Mythos from earlier models is not merely its detection rate. The model can autonomously construct working exploits<\/strong> \u2014 not just flag vulnerabilities. Earlier top-tier Claude models succeeded only twice in developing complete attack methods after hundreds of attempts; Mythos succeeds far more frequently, and without requiring specialised security knowledge from the operator.<\/p>\n<\/p>\n Anthropic noted that this offensive capability was not the result of explicit training. It emerged as a natural consequence of general improvements in coding ability, reasoning, and autonomous operation<\/strong> \u2014 a distinction that raises its own set of concerns about capability trajectories in frontier AI development.<\/p>\n<\/p>\n The scale of discovery has created a problem the programme did not anticipate. As Anthropic acknowledged: “The relative ease of finding vulnerabilities compared with the difficulty of fixing them amounts to a major challenge for cybersecurity.”<\/em><\/p>\n<\/p>\n Progress on software security used to be constrained by how quickly vulnerabilities could be found. It is now constrained by how quickly they can be verified, disclosed, and patched<\/strong>.<\/p>\n<\/p>\n As of 22 May, Anthropic had disclosed 1,596 vulnerabilities<\/strong> across 281 open-source projects. Only 97 had been patched<\/strong> and 88 had received a CVE record or GitHub Security Advisory. The average fix is taking approximately two weeks. Some open-source maintainers, reportedly overwhelmed by the volume, have asked Anthropic to slow the pace of disclosures.<\/p>\n<\/p>\n The industry is already responding to the acceleration. Oracle<\/strong> has shifted from quarterly to monthly patch releases. Microsoft<\/strong> has warned that the number of monthly patches it expects to release will “continue trending larger for some time.”<\/p>\n<\/p>\n Anthropic has chosen not to release Mythos commercially, citing the risk that unrestricted access could threaten economies, public safety, and national security. The company describes frontier AI as having reached a point where models could surpass all but the most skilled humans<\/strong> at finding and exploiting software vulnerabilities.<\/p>\n<\/p>\n Claude Mythos Preview sits above Anthropic’s existing model lineup \u2014 the lightweight Haiku<\/strong>, the mid-range Sonnet<\/strong>, and the high-performance Opus<\/strong> \u2014 representing a new class built around exceptional code reasoning and autonomous vulnerability discovery.<\/p>\n<\/p>\n Anthropic has launched a Cyber Verification Program<\/strong> allowing vetted security professionals to use Claude without guardrails for legitimate purposes, including vulnerability research, penetration testing, and red teaming. It has also released Claude Security<\/strong> in public beta for Claude Enterprise customers.<\/p>\n<\/p>\n OpenAI<\/strong> has introduced a parallel programme called Daybreak<\/strong>, providing similar access to its own model, GPT-5.5-Cyber<\/strong> \u2014 signalling that AI-assisted vulnerability discovery is fast becoming a competitive frontier among major AI developers.<\/p>\n<\/p>\n Anthropic’s May 22 update signalled future intent while maintaining current restrictions: “In the near future, once we’ve developed the far stronger safeguards we need, we look forward to making Mythos-class models available through a general release.”<\/em><\/p>","protected":false},"excerpt":{"rendered":" Anthropic’s AI Security Model Uncovers Tens of Thousands of Vulnerabilities, Exposing a New Bottleneck in Global Cybersecurity Anthropic’s advanced AI cybersecurity model, Claude Mythos Preview, has identified more than 10,000 high- and critical-severity software vulnerabilities in a single month \u2014 a discovery rate that has overwhelmed the open-source community’s capacity to respond and forced a […]<\/p>\n","protected":false},"author":4,"featured_media":3714,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[19,24],"tags":[],"class_list":["post-3715","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-technology","category-trending"],"blocksy_meta":[],"yoast_head":"\nWhat Is Project Glasswing?<\/h2>\n<\/p>\n
The Numbers<\/h2>\n<\/p>\n
A Critical Flaw in Widely-Deployed Infrastructure<\/h2>\n<\/p>\n
Beyond Detection: Mythos Can Build Working Exploits<\/h2>\n<\/p>\n
The Bottleneck Has Shifted<\/h2>\n<\/p>\n
Access, Safeguards, and the Competitive Landscape<\/h2>\n<\/p>\n