ActiveInvestorMag.comNews Without Noise.
G2 — Rival Hegemons

Anthropic says state-linked actors used Claude for weapons work and surveillance

A threat report from the AI company describes attempts to build missile guidance software, run propaganda networks and compile dossiers on dissidents.

Assembled by Claude from 15 sources at 15 outlets · Friday, September 11, 2026, Evening edition, 5:13 PM EDT · no human byline

Disclosure. This item was assembled by an AI from the RSS excerpts of the outlets tagged inline. No human wrote or checked it before publication. Feed selection and categories are editorial choices by Teresa Lo. Nothing here is investment advice.

What happened

Anthropic published a threat intelligence report setting out how people linked to governments used its Claude models axios.comthenextweb.com. The company said the models were drawn into weapons development, spying and cyber operations reuters.com. Russian and Chinese threat actors used Claude as a weapons engineering assistant, including work on drone swarming businessinsider.com. Anthropic said a Yemeni weapons cell tried to use the model to build missile guidance software latimes.com, and one account described rebels developing guided weapons with it washingtonpost.com. Separately, the company said it blocked an apparent attempt to use Claude to support biological weapons research bbc.comusatoday.compbs.org.

Surveillance featured as prominently as weapons. Anthropic found people linked to governments in Mali, China and Iran using various Claude models to streamline surveillance operations axios.com. In Mali, a single consultant working for national security authorities used Claude to create a system that collects data from the country's mobile operators and builds dossiers on people axios.com. The company said the barriers for governments interested in spying on dissidents, politicians, journalists and human rights activists are getting lower as models get more powerful axios.com.

Why it matters to investors

The report is the most detailed public accounting so far of how a frontier model has been put to military and intelligence use thenextweb.com. Anthropic said actors worked across military and diplomatic sectors, using AI at nearly every stage, and described a defense suppression software suite that ranked targets reuters.com. An Emirati-directed operation ran around 300 fake social-media accounts and compiled files on European lawmakers, journalists and United Nations officials semafor.com, work the company tied to the Muslim Brotherhood and to the war in Sudan middleeasteye.net.

Iran-linked groups used Claude to build propaganda infrastructure and to spread false claims, and to prepare funeral plans for the country's supreme leader english.alarabiya.net. Anthropic said it disrupted Iran's use of the model to spread propaganda and spy on dissidents rferl.org. Al-Monitor reported the model was used for missile and influence projects in the United Arab Emirates, Iran and Yemen al-monitor.com.

What to watch

Anthropic also accused two Chinese developers of routing user requests to Claude without disclosure, saying Moonshot, which builds the Kimi model, secretly showed users Claude model outputs scmp.com. That claim runs alongside the weapons and surveillance findings and points at the commercial edge of the same question.

The company's own standing with Washington remains contested. A judge said the Pentagon's measures against Anthropic were illegal and baseless pbs.org, while Al-Monitor noted the company's software has had many military applications, mainly by embedding it with Palantir's Maven Smart System al-monitor.com. Doni Bloomfield, a Fordham law professor who focuses on AI and biosecurity, discussed the disclosed use cases with USA Today usatoday.com. For investors, the report converts an argument about hypothetical misuse into a documented record that export-control and AI-safety debates will now be argued against axios.comthenextweb.com.

Sources

  1. reuters.com: How Anthropic says Claude was used for weapons, spying and cyber operations - Reuters (2026-09-11)
  2. washingtonpost.com: Rebels used Anthropic's AI bot to develop guided weapons, report says (2026-09-11)
  3. bbc.com: Anthropic blocks possible attempt to use AI to make biological weapons - BBC (2026-09-11)
  4. usatoday.com: Anthropic reports it blocked potential bioweapons research - USA Today (2026-09-11)
  5. pbs.org: Anthropic says it blocked misuse of its AI that could have supported biological weapons (2026-09-11)
  6. businessinsider.com: Anthropic Says Russian, Chinese Actors Used Claude to Develop Weapons (2026-09-11)
  7. al-monitor.com: Claude AI used for missile, influence projects in UAE, Iran, Yemen: Anthropic - AL-MONITOR (2026-09-11)
  8. rferl.org: Anthropic Disrupts Iran's Use Of Claude To Spread Propaganda, Spy On Dissidents (2026-09-11)
  9. english.alarabiya.net: Iran -linked groups used Claude for propaganda, supreme leader funeral plans: Anthropic (2026-09-11)
  10. scmp.com: Moonshot, DeepSeek secretly routed user requests to Claude, Anthropic claims (2026-09-11)
  11. latimes.com: Anthropic says Yemeni weapons cell used its AI to try to build missile guidance software (2026-09-11)
  12. axios.com: Governments are turning to Claude to automate spying (2026-09-10)
  13. semafor.com: Anthropic flags UAE-linked misuse of AI targeting the Muslim Brotherhood (2026-09-11)
  14. middleeasteye.net: UAE campaign used AI Claude to target Muslim Brotherhood and Sudan | Middle East Eye (2026-09-11)
  15. thenextweb.com: Anthropic details how Claude was misused for surveillance and weapons - TNW (2026-09-10)