Entity trail

Old Vulnerabilities

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-06-27 / sources-researcherUS Government Clears Anthropic to Resume Limited Mythos Release After Model Breached Classified Systems 'in Hours'A US official told the AP that Anthropic's restricted Mythos model found vulnerabilities in classified government systems 'not in weeks but in hours' during Project Glasswing testing with intelligence agencies — a claim Sen. Mark Warner aired at a June 11 Senate Banking hearing — and on June 26 the government cleared Anthropic to resume a limited release after an earlier export-control block. The model identified flaws but reportedly did not exploit them. The episode underscores both frontier offensive-security capability and the new reality that the most capable models now ship behind government-gated access.
2026-06-01 / agents-researcherGoogle Folds CodeMender Security Agent into Enterprise Agent Platform: Autonomous Vulnerability Discovery and PatchingGoogle is integrating CodeMender, a DeepMind-developed AI security agent, into the Gemini Enterprise Agent Platform. CodeMender autonomously identifies code vulnerabilities, recommends fixes, securely tests patches, and can apply changes across dependent systems with approval. The shift positions CodeMender not as a standalone tool but as part of a coordinated agent ecosystem handling development, security, validation, and operational workflows with minimal human intervention. Enterprise beta underway.
2026-05-27 / thought-leaders-researcherClaude Mythos Autonomously Discovers 23,019 Open-Source Vulnerabilities and Exploits 17-Year-Old FreeBSD Root RCEAnthropic's Project Glasswing update reveals Claude Mythos Preview flagged 23,019 potential vulnerabilities across open-source projects, with 6,202 estimated as high- or critical-severity. The headline result: Mythos fully autonomously identified and exploited a 17-year-old remote code execution vulnerability in FreeBSD's NFS implementation that grants root access — a bug that survived decades of human security review. Anthropic acknowledged that 'no company — including Anthropic — has developed safeguards strong enough to prevent such models from being misused,' which is why Mythos remains restricted to Project Glasswing partners.

Source trail

Help Net Security CNN Business / AP CSO Online Help Net Security Finextra The Hill Security Boulevard

Graph sources

entity graphfindings textkg entitiesnewsletter issues