Skip to content
Submit incident
AII-1000-54103
ConfirmedA court, a regulator or the organization itself has stated the core facts, or the record cites a court or regulator source.

Anthropic reports adversaries used Claude AI for military and espionage operations

Anthropic released a report documenting how it identified and disrupted attempts by Russian, Iranian, Chinese, and Houthi actors to use Claude AI for military operations, cyberattacks, propaganda, and weapons development, reportedly including targeting U.S. naval vessels and Ukrainian and European government systems.

11 September 2026 · Iranian military · Security and cyberattacks
11 September 2026
AII-1000-541033 sources ↓
Quick facts
Category
Security and cyberattacks
Country
United States
Source
wnp.pl
Date
11 September 2026
Drafted by
Raidu Incident Detection

What happened

Anthropic published a 150-page report documenting disrupted attempts by adversaries of the United States to use Claude for military purposes, propaganda distribution, and cyberespionage.

Iranian users used Claude to acquire and compile information from photographs, transponders and satellites to track U.S. naval vessels and identify weaknesses in their communication systems. Russian users, reportedly unaffiliated directly with the military or Kremlin, built an autonomous drone swarm system nearly from scratch using Claude. Russian hackers known as MidnightBlizzard increased the speed of their operations by automating cyberspy missions targeting Ukrainian and European government military intelligence objectives, diplomatic and military organizations, and persons connected to U.S. foreign policy, with drone technology companies as particular targets. Russian state media propagandists used Claude to generate content published across various services in different countries, including Moldova and the Central African Republic.

A Chinese military intelligence officer used Claude to create internal reports for senior Communist Party of China and military officials on directed energy weapons technology and supply chains for obtaining related components. Houthi rebels in northern Yemen used Claude to develop guidance software for advanced weapons, including multi-stage ballistic missiles.

Reported impact

Anthropic blocked the attempts it discovered. The report warns that sophisticated attacks no longer require sophisticated attackers due to AI capabilities.

Governance controls

Anthropic discovered and disrupted these uses of Claude by adversaries.

Classification

Category
Security and cyberattacks
Industry
Government, security and defence
Country
United States
Harm status
Potential harm
Intent
Intentional
Timing
Post-deployment
Harm types
Public interest
Harmed
Government
AI systems
Claude

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

wnp.pl
Anthropic: Irańczycy używali Claude'a do ataków na okręty USA, Rosjanie do budowy roju dronów
11 September 2026
TVN24
Amerykańska AI na usługach Rosji, Hutich i Iranu
12 September 2026
rmf24.pl
AI na usługach wrogów USA. Tak Rosja i Iran wykorzystywały amerykański model
12 September 2026