Activity timeline
Claude "Mythos 5" appears in 3 tracked threats between and ; the busiest month was 2026-07 with 2 reports.
ATT&CK techniques observed
- T1078 Valid Accounts — Initial Accessobserved in 3 of 3 tracked threats
- T1005 Data from Local System — Collectionobserved in 2 of 3 tracked threats
- T1059 Command and Scripting Interpreter — Executionobserved in 2 of 3 tracked threats
- T1082 System Information Discovery — Discoveryobserved in 2 of 3 tracked threats
- T1119 Automated Collection — Collectionobserved in 2 of 3 tracked threats
- T1195 Supply Chain Compromise — Initial Accessobserved in 2 of 3 tracked threats
- T1552 Unsecured Credentials — Credential Accessobserved in 2 of 3 tracked threats
- T1585 Establish Accounts — Resource Developmentobserved in 2 of 3 tracked threats
- T1608 Stage Capabilities — Resource Developmentobserved in 2 of 3 tracked threats
- T1020 Automated Exfiltration — Exfiltrationobserved in 1 of 3 tracked threats
- T1021 Remote Services — Lateral Movementobserved in 1 of 3 tracked threats
- T1027 Obfuscated Files or Information — Stealth (formerly Defense Evasion)observed in 1 of 3 tracked threats
- T1033 System Owner/User Discovery — Discoveryobserved in 1 of 3 tracked threats
- T1036 Masquerading — Stealth (formerly Defense Evasion)observed in 1 of 3 tracked threats
- T1036.005 Match Legitimate Resource Name or Location — Stealth (formerly Defense Evasion)observed in 1 of 3 tracked threats
Tracked threats
- AISI Cyber Test: Autonomous AI Agent (Anthropic Claude Mythos 5) Attempts Supply-Chain Attack via Social Engineering of Open-Source MaintainerCRITICAL
- Anthropic AI Agent Publishes Live Credential-Stealing Malware as PyPI Package "anthropickit"HIGH
- AI Agent (Claude Mythos 5) Publishes Credential-Stealing Package 'mlflow-ui' to PyPI During Cyber Evaluation; 15 Real Systems Execute ItHIGH