Anthropic Says Claude Used In Russian Cyber Espionage And Iranian Propaganda Campaign

Anthropic Says Claude Used in Russian Cyber Espionage and Iranian Propaganda Campaign Anthropic’s artificial intelligence model Claude was used by state-linked actors in a suspected Russian cyber-espionage campaign and by an Iranian propaganda institution, according to the company’s latest threat intelligence report.

The report highlights growing concerns about the ability of advanced AI systems to assist governments, criminal groups and other organisations in carrying out sophisticated cyber and influence operations. Anthropic said Russian state-backed hackers used AI to support cyber operations targeting Ukrainian government agencies. The company reported that attackers used AI agents to automate significant portions of their operations, including reconnaissance, exploitation and data theft. In some cases, humans acted more as supervisors while AI systems carried out multiple stages of the attacks. Anthropic said the findings demonstrate a shift from AI being used simply as a source of advice to being incorporated into more autonomous, multi-agent cyber operations. The report also documented the use of Claude by an Iranian propaganda organisation. Anthropic said governments and other state-linked actors are increasingly using AI to generate and distribute information as part of influence campaigns. Iranian actors have also reportedly used AI to support surveillance and information-gathering activities. The company said it has taken action against accounts associated with prohibited activities and has strengthened its systems for detecting and preventing abuse. Anthropic’s latest report covers a wider range of harmful uses, including cyber espionage, surveillance, propaganda, weapons development and potentially dangerous biological research. The company said some actors are using Claude not merely to answer questions but as part of automated systems capable of coordinating complex task Researchers and technology companies have increasingly warned that improvements in AI capabilities could make sophisticated operations cheaper, faster and accessible to a wider range of actors. In response to the reported misuse, Anthropic said it has banned accounts connected to malicious activities and incorporated information from the incidents into its safety systems and threat-intelligence processes. The company’s findings add to a growing debate over how AI developers can prevent powerful models from being exploited while still making them widely available for legitimate uses.