Group in Yemen used Claude AI for missile projects, Anthropic says

A Yemeni group attempted to utilize Anthropic's Claude chatbot for creating guided rockets and missiles, according to a recent report by the US-based AI firm. In an effort to curb illegal uses of Claude across multiple nations, Anthropic disclosed that it had uncovered a "cell of threat actors based in northern Yemen running three weapons development programmes".

These nefarious activities included a guided rocket utilizing a low-cost flight computer with final-phase guidance, a multi-stage ballistic missile with a range goal exceeding 2,000 kilometres, and a multi-variant missile with a hypersonic glide vehicle variant.

The aforementioned cell allegedly employed Claude to develop guidance, navigation, and control software for these airborne weapons, which would otherwise necessitate an entire team of engineers. While there is no definitive evidence that the actors successfully deployed an operational device, they did conduct a field test for a guided rocket, which unfortunately failed within hours.

Subsequently, the actors returned to Claude to investigate the cause of the failure. Anthropic promptly banned the responsible accounts upon detecting the illicit activity. Although the company did not disclose the identity of the actors, it is plausible that they are affiliated with the Houthis, who wield control over a significant portion of northern Yemen, including the capital, Sanaa.

This development occurs amidst a surge in Houthi offensive activities throughout Yemen, with the group making strides towards capturing the strategic Bab al-Mandab strait, a crucial waterway connecting Europe to Asia. On Friday, the Houthis claimed control of a Red Sea island situated in the Bab al-Mandab shipping lane, effectively completing their effort to seize control of the area.

In addition to the weapons development scheme, Anthropic also disrupted attempts to utilize its Claude models for the creation of biological weapons and a suspected Russian-linked cyber espionage campaign targeting Ukraine. The AI firm implicated Chinese rivals in hacking endeavors aimed at extracting capabilities from Claude. Furthermore, two Anthropic researchers recently cautioned that rapid advancements in artificial intelligence could potentially trigger the extinction of the human race in the near future.