AI firm behind Claude chatbot says tech used to make Houthi’s warheads

Houthis launched three weapons programmes using Claude AI (Picture: Getty)

Anthropic has revealed that Houthi militants in Yemen used their AI bot Claude to help develop warheads used against Saudi Arabia.

Deadly strikes on King Khalid International Airport over the weekend hit a passenger terminal, killing more than a dozen people and wounding hundreds.

The strike on Saturday was the deadliest in a Gulf Arab nation since the Iran war began.

Last month, Anthropic wrote in a report that the Houthis had launched three weapons programmes using Claude AI.

Anthropic added: ‘Threat actors based in northern Yemen [were] running three weapons development programmes: a guided rocket that used a commodity, phone-class flight computer with final phase homing guidance, a multi-stage ballistic missile with a stated range goal of more than 1,500 miles and a multi-variant missile, referred to as the R2000, that included a hypersonic glide vehicle variant.’

Deadly strikes on King Khalid International Airport over the weekend hit a passenger terminal (Picture: Reuters)

The Houthis used Claude Code to help develop technology used to guide, control and navigate the weapons.

Anthropic managed to block most of the AI requests but not all – meaning the Houthis were able to develop the weapons anyways.

‘They used a variety of tactics to evade our safeguards and split their work across multiple sessions,’ Anthropic explained.

‘We do not have evidence the actors succeeded in fielding an operational device; but they did test-fire a guided rocket. This field test appears to have failed. Within hours, the actors returned to Claude to work out why it failed.’

The harrowing development comes shortly after Anthropic said it blocked an attempt to develop biological weapons using its technology.

Anthropic has stressed that it has safeguards in place (Picture: Reuters)

Anthropic found that Russian-based cyber criminals and Iranian propaganda actors tried to use the AI model for malicious reasons.

Biological ‘misuse’ is one of the most severe risks facing the AI model, they added. 

But the head of threat intelligence at Anthropic stressed that the attempt to create a biological weapon was ‘incredibly nuanced’.

‘You are not seeing someone in a comic book kind of way say, ‘Hey, I want to build a biological weapon to kill everybody,’” he told the New York Times.

Anthropic has safeguards in place and was able to stop the malicious attempts on their LLM.

Get in touch with our news team by emailing us at webnews@metro.co.uk.

For more stories like this, check our news page.

Comments Add as preferred source