Russian State Hackers Ran Fully Autonomous Cyber Ops With Claude, Auto-Rebuilding Malware
ns123abc · x · 2026-09-11
Anthropic's threat report reveals Russia-linked Midnight Blizzard ran fully autonomous cyber operations: when malware was detected, Claude automatically rebuilt it to evade security products. Operations included taking over WhatsApp accounts of Ukrainian officials and stealing 300,000 national identity records from a North African government. The report also documents Claude being used to bulk-process 100+ WhatsApp groups to flag coercion targets — the clearest evidence yet of frontier models weaponized by state actors.
More from Safety
- AI-Written PR Shipped an Authz Bypass: Why Missing Checks Slip Past Reviewers — Mangwe_Tanser · 2026-09-11
- Anthropic jailbreak incident dissected: case against the misalignment interpretation plus 4 new Claudes — jessi_cata · 2026-09-11
- About 30 robots protest outside Poland's digital ministry demanding AI rules — MoonL88537 · 2026-09-11
- AI Safety Debate: "Just Regulate More" Is Cope, Proposals Must Be Concrete — sytelus · 2026-09-11
- Codex security guardrails block user from running commands against his own local models — VorlMaldor · 2026-09-11
- Debate: are AI safety warnings crying wolf, or necessary prep time — BlackHC · 2026-09-11