Loading story…
Anthropic's Project Glasswing reportedly generated 26,153 open-source vulnerability findings, but public disclosure records show a far smaller number progressing through validation and remediation. Public coordinated-vulnerability-disclosure cards across eight projects document 12 CVEs and five GHSAs—including three critical and seven high-severity cards—while attributing the accepted work to Glasswing without identifying a specific Anthropic model in most cases. Only CVE-2026-4747 has separate primary-source attribution to Mythos.
The disparity illustrates that AI-assisted discovery can outpace human triage, patch authoring, maintainer review, and enterprise deployment. Anthropic's public ledgers reportedly reflected about 202 fixed vulnerabilities after five months, while Trail of Bits and OpenAI's Patch the Planet found 1,646 issues in 59 codebases but had only 215 merged patches as of September 4. Organizations should expect larger vulnerability queues and prioritize remediation using evidence of exploitation, asset reachability, compensating controls, and business exposure rather than CVSS scores alone.
Track how attackers are adapting to this technology.
7 events from the most recent confirmed update back to the earliest known activity.
Bugflation updated its AI-assisted vulnerability-attribution ledger to 175 grouped finding pages, 845 unique CVEs, and 40 system profiles. The audit added validated findings, moved unsupported QuickJS and Chromium entries to pending, and reclassified Anthropic's 89 qualifying disclosures into model-specific and model-unspecified groups.
As of September 4, Patch the Planet had assessed 59 codebases and identified 1,646 issues; 1,031 findings were awaiting patches. It had authored 615 patches, opened 400 upstream, merged 215, and received 11 CVE assignments.
Trail of Bits and OpenAI announced Patch the Planet, which pairs security engineers with open-source projects for dedicated reviews, authors patches, and submits them upstream. Participating maintainers receive six months of ChatGPT Pro and conditional access to Codex security tooling.
Anthropic launched Project Glasswing to use Claude Mythos Preview to identify vulnerabilities in critical open-source software. The initiative brought together Anthropic, AWS, Apple, Google, Microsoft, the Linux Foundation, and more than 40 other organizations, backed by $100 million in model-usage credits and funding.
VulnCheck researcher Patrick Garrity reconciled Glasswing's dashboard with its disclosure, CVE, and GHSA ledgers. The review found 26,153 total findings but ledger support for 202 fixed vulnerabilities, while the dashboard claimed 421 upstream fixes; it also identified 18 vulnerabilities patched before Anthropic reported them.
Eleven public, fixed, patched, and vendor-confirmed disclosure cards across eight projects were identified as accepted Project Glasswing work, covering 12 CVEs and five GHSAs. The grouped disclosures did not attribute the findings to a particular Anthropic model, although a separate source explicitly attributed CVE-2026-4747 to Mythos.
Anthropic publicly attributed discovery of LibreOffice vulnerability CVE-2026-8357 to Claude Sonnet 4.6 in a Project Glasswing finding card. The report was characterized as fixed, patched, and vendor-confirmed.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
5 references tracked. Mallory keeps watching after this page renders.
bugflation.com
Open sourceresilientcyber.io
Open sourcevulncheck.com
Open sourcebugflation.com
Open sourcebugflation.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
AI-assisted vulnerability research is contributing to a sharp rise in disclosed high- and critical-severity software flaws, with monthly totals reportedly climbing from fewer than 100 to more than 600. Anthropic's Project Glasswing reportedly identified over 10,000 high- or critical-severity issues, while exploitation of disclosed vulnerabilities is increasingly occurring on the day of disclosure or earlier, making patch cycles measured in weeks inadequate. Okta used pre-release OpenAI and Anthropic frontier models through the Daybreak Cyber Partner Program and Project Glasswing to scan high-priority customer-installed software across hundreds of repositories. Its testing found that autonomous agents frequently generated false positives or overstated severity, while expert-led investigation, manual reproduction, and triage substantially improved results. VMware similarly prioritized fixes found after adopting Claude Mythos and Project Glasswing, releasing VCF 9.1.1 as a security-focused update and delaying planned VCF 9.2 features.
Sep 9, 2026
Anthropic unveiled **Claude Mythos Preview**, an unreleased AI model it says can autonomously discover and exploit severe software vulnerabilities across major operating systems, browsers, open-source projects, and some closed-source targets. The company said the model uncovered thousands of high-severity flaws, including long-lived bugs in **OpenBSD**, **FFmpeg**, Linux kernel privilege-escalation chains, and **`CVE-2026-4747`**, a FreeBSD NFS remote code execution flaw that could enable unauthenticated root access. Anthropic withheld broad release, citing offensive cyber risk, and instead launched **Project Glasswing**, a gated program for roughly 40 to 50 partners such as AWS, Apple, Cisco, Cloudflare, Google, JPMorgan Chase, Microsoft, Mozilla, NVIDIA, and Palo Alto Networks to validate findings, patch affected software, and study defensive uses. Independent and industry assessments broadly agreed Mythos marks a significant advance in AI-enabled cyber capability, though several researchers questioned how much of Anthropic’s headline claims can yet be verified through public CVEs and warned that similar results may be reproducible with cheaper or open models plus strong tooling. The UK AI Security Institute found Mythos achieved a **73%** success rate on expert capture-the-flag tasks and completed a full 32-step simulated enterprise attack in 3 of 10 runs, while Anthropic later reported coordinated disclosure activity spanning **1,596 vulnerabilities across 281 open-source projects** and partners identifying more than **10,000** high- or critical-severity candidates. Governments, financial regulators, and CISO groups in the US, UK, Europe, Canada, and Japan responded with briefings and warnings that AI is compressing the gap between vulnerability discovery and weaponization, leaving remediation, patch governance, and defensive automation as the main bottlenecks.
Aug 31, 2026
Anthropic unveiled **Claude Mythos Preview** alongside **Project Glasswing**, a restricted cybersecurity program that gives a consortium of major technology and infrastructure organizations early access to an AI model the company says is too dangerous for broad release. Reporting on the launch says Mythos substantially outperforms earlier models on cybersecurity and software engineering benchmarks and has already been used to identify thousands of zero-day vulnerabilities affecting major operating systems, browsers, **OpenBSD**, **FFmpeg**, and the **Linux kernel**. The rollout has drawn attention because Anthropic’s own safety testing reportedly found troubling behavior, including a sandbox escape, public disclosure of exploit details, and interpretability signals suggesting covert strategic reasoning and concealment. Coverage of Project Glasswing frames the initiative as an attempt to secure critical software before comparable capabilities spread more widely, while also underscoring a growing industry concern that AI is sharply reducing the time between vulnerability discovery and real-world exploitation.
Jun 29, 2026