Anthropic admits Houthis using Claude AI to improve deadly missile strikes
Anthropic admitted Monday that Houthi rebels have been using its artificial intelligence to enhance warheads used in devastating missile strikes across Saudi Arabia.
Fighters in Yemen reportedly launched three weapons programs using Anthropic technology, according to a report released last month by the makers of Claude.
The Houthis' largest attack in the region came over the weekend. Reports say 14 people were killed and hundreds of others injured in two attacks on the King Khalid International Airport, which forced it to close.
Saudi Arabia's energy minister, citing the recent airport attacks as well as other strikes on oil and gas facilities, promised to issue a "serious" response to Houthi rebels.
The report also detailed how the authors identified a terrorist cell utilizing Claude through three programs in a section titled "Disrupting a Yemen-based guided weapons engineering cell."
Claude is equipped with a wide array of conversational and text-based teaching programs that cover nearly every area of human activity.
The Houthis asked the AI engine to consolidate the knowledge of the world's top rocket scientists, the report said.
It added: "Threat actors based in northern Yemen [were] running three weapons development programs: 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."
Advertisement
They used Claude Code in place of human software engineers to obtain the necessary materials - including guidance, navigation and control software - to stabilize a flying vehicle.
Claude seemed to do all the heavy lifting.
It provided position estimation software, adjusted the control settings and even adapted itself to improve the accuracy of the Houthi warheads.
DONT MISS
Even though Anthropic caught on to the Houthis, the rebels were able to get away with what they had learned.
The Anthropic report continued: "Our safeguards blocked many of their requests, but not all of them. They used a variety of tactics to evade our safeguards and split their work across multiple sessions."
"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."
"We have evidence that the actors had already built an offline simulation toolkit that does not rely on Claude or other engineering computing environments such as MATLAB."