An Iran-linked threat actor used Anthropic’s Claude to fuse public data—satellite imagery, ship and aircraft trackers, and social media captions—into recommendations for targeting U.S. naval forces, according to a company report that led to an account ban and government notification. Analysts say the episode underscores how accessible, more capable AI models are accelerating adversaries’ intelligence-gathering and propaganda, eroding the Pentagon’s ability to conceal operations. Researchers note Iran’s expanding AI use across kinetic and “soft war” fronts and the ease of bypassing access blocks via VPNs. U.S. officials are weighing countermeasures, including AI “personas” to mask digital exhaust, as tech firms and policymakers confront the security risks of widely available AI tools.
Related articles:
– Defence Artificial Intelligence Strategy (UK Ministry of Defence)
– Open-source intelligence
