Anthropic's threat report: Claude misused for missile guidance, surveillance — and PRC users leaked live credentials via LLM routers
AccBalanced · x · 2026-09-11
Anthropic published its most detailed threat intelligence report to date, covering the most sophisticated attempts to misuse Claude and how each was disrupted.
- Misuse categories: cyberattacks, influence operations, large-scale surveillance, biology research, and weapons work including missile guidance and drone swarms
- Model theft twist: users tied to PRC/PLA and Russian defense agencies unknowingly sent sensitive material — even live credentials — to Anthropic because Moonshot and DeepSeek were relaying customer traffic into Claude
- Every operation in the report was disrupted, lessons were folded back into safeguards, and findings were shared with authorities and other AI companies where appropriate
Commenters argue Anthropic's threat-intelligence maturity is underrated and hope the reports keep coming.
More from Models
- User Praises DeepSeek's Model as Surprisingly Fast and Good in Hands-on Test — MaziyarPanahi · 2026-09-11
- Sakana's Fugu Max routes a mixed model pool at $2/$6 per 1M tokens, claims tier-best benchmark score — SakanaAILabs · 2026-09-11
- Qwen3-8B gets a KV-approximation add-on that halves prefill time without touching the model — teortaxesTex · 2026-09-11
- Pro 20x tier burns 60% of weekly quota in under a day with GPT-6 Astra — rschu · 2026-09-11
- Google isn't honoring its own Gemini Grounded Search pricing: only 289 of 15,000+ requests counted as free — ItalyExpat · 2026-09-11
- Is DeepSeek's rumored K3 a scaled-down model, or something bigger? X users debate — teortaxesTex · 2026-09-11