Anthropic said it suspended the accounts in question as soon as it discovered the illegal activity. File | Photo credit: Reuters
A group in Yemen has attempted to use chatbot Anthropic Claude to create guided missiles and missiles as Iran-backed Houthi rebels wage an offensive in the country, a US firm said.
In a report published on Thursday (September 10, 2026) on how Anthropic thwarted Claude’s illegal use in several countries, it said it had “identified a malicious cell based in northern Yemen with three weapons development programs.”
They were “a guided missile that used a conventional telephone-class on-board computer with terminal phase homing; a multi-stage ballistic missile with a stated range of more than 2,000 kilometers (1,250 mi); and a multi-variant missile… which included a hypersonic glide vehicle variant.”
The cell used Claude to develop guidance, navigation, and airborne weapons control software that would otherwise have required a team of engineers.
“We have no evidence that the actors were able to use a functioning device; but they tested a guided missile,” Antropik said.
“This field test appears to have failed: a few hours later the actors returned to Claude to find out why it failed.”
Anthropic said it suspended the accounts in question as soon as it discovered the illegal activity.
While the company did not name the actors involved, it likely involves the Houthis, who control much of northern Yemen, including the capital Sanaa.
The discovery comes as Iran-backed militants have launched a large-scale offensive across the country towards the strategic Bab al-Mandab strait, which connects Europe to Asia.
Anthropic Bullying: Artificial Intelligence Firm Anthropic Against the US Government
On Friday (September 11, 2026), the Houthis took control of an island in the Red Sea in the middle of the Bab al-Mandeb shipping route, ending their push to control the area.
Published – September 11, 2026 05:56 pm EST.