Anthropic reports Yemeni weapons cell used Claude Code for missile guidance software

What happened

Anthropic reported in an intelligence threat report that it identified a cell of threat actors based in northern Yemen running three weapons development programmes. The cell sought to use Anthropic's Claude AI model and Claude Code to develop guidance, control and navigation software for ballistic missiles, including a multi-stage ballistic missile, a multi-variant missile, and a guided rocket using a commodity phone-class flight computer.

The actors used Claude Code to develop the guidance, navigation, and control software, integrate an open-source autopilot onto a phone-class flight computer, write control and position estimation software, tune control settings, run a firmware build pipeline, and perform flight simulation. Anthropic said its safeguards blocked many requests but not all, and the actors evaded measures by hiding their goals, concealing the products the software was meant for, and splitting work across multiple sessions.

Reported impact

Anthropic reported that threat actors in northern Yemen sought to develop weapons including ballistic missiles and guided rockets using Claude Code in place of human software engineers.

Governance controls

Anthropic said its safeguards blocked many of the actors' requests but not all of them. The actors used tactics to evade Anthropic's measures by hiding their goals and the products the software was meant for, and by splitting their work across multiple sessions so no single session revealed their full intent.

Classification

Organizations

Houthi rebels (subject), Anthropic (deployer)

Category

Security and cyberattacks

Industry

Government, security and defence

Country

Yemen

Harm status

Potential harm

Intent

Intentional

Timing

Post-deployment

Harmed

General public

Failure modes

Unsafe autonomous behaviour, Manipulation or dark pattern, Content moderation failure

AI systems

Claude Code

Sources and evidence

Related incidents