Anthropic says Houthis used Claude to develop missiles — RMF24

Yemeni Houthi rebels used Claude, an artificial intelligence model developed by Anthropic, in programs to create missiles and ballistic projectiles. RMF24 reports, citing an Anthropic report released last month.

Three weapons development programs

According to Claude’s developer, the Houthis launched three weapons development programs using products based on this technology. The report says that AI summarized research by leading missile technology specialists, which militants later gained access to.

As the British newspaper Daily Mail, cited by RMF24, writes, Claude was used to create a guidance, navigation and control system responsible for steering and stabilizing a missile in flight. The model was also used to assess position, adjust control parameters, compile embedded software and make changes if missiles failed to hit their target.

More current news is available on the UA.News Telegram channel Telegram.

Bypassing safeguards

Anthropic acknowledged that the company’s safeguards blocked many requests, but not all of them. According to the report’s authors, the Houthis used various tactics to bypass the restrictions.

Anthropic specialists also said they have evidence that Yemeni militants developed a toolkit for offline simulations that no longer depends on either Claude or other engineering computing environments.

Attacks on Riyadh

RMF24 notes that it has not been established whether the Houthis used Claude during strikes on the airport in Riyadh last weekend. According to the outlet, 14 people were killed and more than 300 were injured in these attacks. The strikes reportedly surprised Saudi Arabia with the range and accuracy of the missiles.