Anthropic said it disrupted a Russia-linked campaign by Midnight Blizzard that used Claude to automate malware evasion and data theft across more than 20 targets between December 2025 and August 2026. The victims included government, defense, diplomatic, intelligence, and drone-supply-chain organizations in Europe, the Middle East, and Asia.
The workflow turned detection into a feedback loop: when security products flagged an implant, AI agents used Claude to test, modify, rebuild, and redeploy it until it slipped past again. In plain terms, the malware kept coming back in a slightly changed form, which makes static signatures and one-time block rules a moving target instead of a finish line.
For defenders, the shift is bigger than one cluster. If your detection stack assumes malware stays mostly stable, this is the case that shows how AI can keep regenerating variants faster than manual triage ages them out, and why repeated evasion now has to be treated as part of the attacker’s operating model.
I’m still reading it, but I wanted to flag this: We identified a cell of threat actors based in northern Yemen running three weapons development programs: a guided rocket that used a commodity phone-class flight computer with final-phase homing guidance; a multi-stage ballistic missile with a stated range goal above 2,000 km; and a multi-variant missile (referred to as the “R2000” set) that included a hypersonic glide vehicle variant...
Anthropic says multiple threat groups, including the financially motivated and state-sponsored espionage groups linked to Russia and China, tried to abuse its Claude AI model for malicious purposes.
Anthropic detected and disrupted a Russia-linked cyber-espionage group that used its AI tool Claude in a hacking campaign targeting more than 20 government, intelligence, diplomatic and defense organizations.