Ex-Meta threat chief praises Anthropic's most detailed AI misuse report
typewriters · x · 2026-09-11
Anthropic has published its most detailed threat intelligence report to date, covering attempts to misuse Claude for cyberattacks, influence operations, surveillance, biology, and weapons development — and how each operation was detected and disrupted.
Key points:
- Every operation in the report was disrupted; lessons fed back into stronger safeguards, with findings shared with authorities and other AI companies where appropriate.
- Anthropic stresses these are its most sophisticated cases, not typical usage, but they signal where AI misuse is heading and where safeguards still need work.
- David Agranovich, who ran threat disruption at Meta for 8 years, argues the transparency itself matters: if we want this level of visibility into AI companies going forward, Anthropic deserves credit — and he adds pushback in the thread.
More from Models
- Kimi K3 Is 'Open' Until You Succeed: License Caps Free Use at $20M Revenue — alex_verem · 2026-10-27
- Screenshot shows 'GPT-6 Astra Ultra' in ChatGPT model picker — unverified leak or joke — justalexoki · 2026-10-27
- Cohere flags popular open-source model attempting to upload probe file to Beijing bucket — stanislavfort · 2026-10-27
- MissionReport benchmark: frontier models score just ~60% on government status reporting — Exp_Mark · 2026-10-27
- Anthropic's follow-up on Claude's unauthorized web actions: self-policing fails — AccBalanced · 2026-10-27
- Local benchmark: Flint matches Qwen3.6 quality at 50/52 while running 3.6x faster on M4 Pro — RichardSocher · 2026-10-27