Daijiworld Media Network – Yemen
Yemen, Sep 11: Iran-backed Houthi militants in Yemen may have used Anthropic’s artificial intelligence chatbot Claude to assist in developing software related to guided missile programmes, according to a threat intelligence report released by the technology company.
The 154-page report detailed alleged attempts by governments, criminal groups and other threat actors to misuse Anthropic's AI models for activities including weapons development, espionage, propaganda and other malicious operations.
Anthropic said a Yemen-based group used Claude Code while working on guidance, navigation and control software for guided rockets and missile programmes. Although the company did not identify the group, details cited in the report were said to point towards the Houthis.
According to Anthropic, the actors were working on three missile programmes, including a guided rocket, a multistage ballistic missile with a stated range of more than 2,000 kilometres, and missile variants involving a hypersonic glide vehicle.
The AI tool was reportedly used to assist with coding, research, software reviews, control systems, simulations and firmware-related work. Anthropic said the operators allegedly used several instances of Claude simultaneously for different tasks.
The company said its safeguards blocked many requests, but the actors allegedly attempted to evade restrictions by concealing their intentions and splitting their work across multiple sessions.
Anthropic said it had no evidence that the group successfully deployed an operational guided weapon. However, it noted that a guided rocket was test-fired but apparently failed, following which the operators allegedly returned to the AI system to troubleshoot the programme.
The company said it identified the activity during an internal investigation, banned accounts linked to the alleged actors and shared relevant information with public- and private-sector partners.
The Yemen case was among six instances highlighted by Anthropic involving alleged attempts to use Claude for conventional weapons development. The report also documented alleged misuse involving cyber operations, surveillance, propaganda and potentially dangerous biological research.
Anthropic said the cases represented some of the most sophisticated misuse it had identified and highlighted growing concerns over the potential use of increasingly powerful AI systems for malicious activities.