Claude AI used in Yemen missile projects, Anthropic reveals

The US artificial intelligence company said it identified a group in northern Yemen that was running three separate weapons development programmes.

Anthropic revealed the findings in a report examining attempts to misuse its Claude chatbot for illegal activities in different countries.

The company said the group used Claude to work on software needed for guidance, navigation and control systems for airborne weapons.

According to Anthropic, this type of work could otherwise require a team of engineers with specialised technical knowledge.

Three missile projects identified

Anthropic said the Yemen-based group was working on three weapons programmes with different designs and capabilities.

One project involved a guided rocket using a commercially available phone-class flight computer and a system designed to improve its final-stage targeting.

Another project involved a multi-stage ballistic missile with a stated range target of more than 2,000 kilometres, or around 1,250 miles.

The third programme involved several missile designs, including a version featuring a hypersonic glide vehicle.

The company did not provide evidence that any of these systems became fully operational.

However, Anthropic said the group did test-fire a guided rocket during its development efforts.

The company said the test appeared to fail, and the people involved returned to Claude within hours to investigate what had gone wrong.

That detail has raised fresh questions about how quickly AI tools can provide technical assistance for highly complex weapons projects.

We're publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them. We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies. These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen. But they’re especially important to discuss, because they show us where AI misuse is headed, where our safeguards work, and where they need to improve. We’re publishing this report so others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop. Read the report: https://t.co/0EJUnYEgfz

— Anthropic (@AnthropicAI) September 10, 2026

Anthropic bans accounts after detection

Anthropic said it banned the accounts connected to the activity after detecting that Claude was being used for illegal purposes.

The company did not publicly identify the individuals or organisation behind the weapons programmes.

However, the report has drawn attention to the Houthis because they control large parts of northern Yemen, including the capital, Sanaa.

The disclosure comes during a period of heightened military activity involving the Houthi movement.

The group has been involved in fighting and has carried out operations around strategically important areas in Yemen.

Anthropic’s findings show how AI systems can potentially be sought out for assistance in areas that companies specifically prohibit.

The company said its monitoring systems helped identify the misuse and allowed it to take action against the accounts.

Houthi activity adds to concerns

The disclosure comes as Houthi forces have been carrying out a wider offensive towards the Bab al-Mandab Strait.

The strategically important waterway links the Red Sea with the Gulf of Aden and serves as a major shipping route between Europe and Asia.

The area has become increasingly important because disruptions around the Red Sea can affect international trade and shipping.

The Houthis also recently claimed control of a Red Sea island located in the middle of the Bab al-Mandab shipping lane.

Their claim came as the group sought to strengthen its position around the strategically important maritime area.

Against this backdrop, Anthropic’s disclosure about weapons development has added another layer to concerns surrounding the use of advanced technology in conflict zones.

Also Read:Pakistani-Canadian Zeeshan Hamid appointed Ontario’s new minister

AI misuse goes beyond missile development

Anthropic said its investigation also uncovered attempts to use Claude for other dangerous activities.

The company reported efforts involving the development of biological weapons, as well as a suspected cyber espionage campaign linked to Russia and targeting Ukraine.

The findings suggest that concerns about AI misuse are no longer limited to one type of technology or one region.

Advanced AI systems can help users process information, write software and solve technical problems, which can also create risks when those abilities are used for harmful purposes.

AI companies have therefore increased their focus on detecting suspicious activity and restricting accounts that violate safety rules.

Anthropic said it continues to monitor how its models are being used and takes action when illegal activity is detected.

Anthropic warns about wider AI risks

The company’s latest report comes shortly after two Anthropic researchers raised broader concerns about the rapid development of artificial intelligence.

The researchers warned that increasingly capable AI systems could eventually create serious risks for humanity.

Their warning included the possibility that unchecked advances in AI could pose an existential threat in the future.

At the same time, companies developing AI systems are facing growing pressure to prevent their technology from being used for weapons, cyberattacks and other harmful activities.

The Yemen case provides a real-world example of the challenge.

Anthropic says its system detected the suspected misuse and the related accounts were banned, but the incident shows how users may attempt to turn general-purpose AI into a tool for specialised weapons work.

The company’s findings are likely to fuel further debate over AI safety, monitoring and the responsibility of technology firms.