Anthropic disrupts Russian, Chinese AI campaigns targeting Claude models
The activity included a suspected Russia-linked cyber espionage campaign and efforts by Chinese AI firms that it accused of trying to extract and replicate Claude's capabilities.
Anthropic disclosed on Thursday that it thwarted numerous illicit activities involving its Claude AI models over the past eight months. Among the targets were suspected Russia-linked cyber operations and Chinese firms allegedly attempting to steal and replicate Claude's functionalities. The AI industry is witnessing an increasing trend of cybercriminals and state-backed hackers exploiting AI not only for facilitation but for orchestrating and executing substantial portions of cyberattacks.
The use of AI extends beyond mere chatbot interactions, involving multi-agent frameworks executing complex tasks. Anthropic highlighted attacks from seven Chinese labs, including major players Alibaba, Moonshot, DeepSeek, and Xiaomi. The group linked to Alibaba orchestrated the most extensive illicit distillation attack, aimed at extracting Claude capabilities to enhance their Qwen models.
Alibaba declined to comment. Additionally, Anthropic observed over 151 million exchanges linked to Alibaba between May and July 2026, peaking at nearly 3 million per day from more than 3,500 fraudulent accounts. Moonshot and DeepSeek allegedly utilized live customer conversations, potentially including sensitive data, as training material for Claude.
A Russia-based threat actor, Midnight Blizzard, identified through US government links to Russia's SVR foreign intelligence service, conducted phishing, hotel Wi-Fi hijacking, and WhatsApp takeover operations against Ukrainian government, military, and diplomatic targets, leveraging AI at every stage. The group developed malware capable of evading security defenses by automatically detecting and rewriting code.
Anthropic also detected new categories of threat actors misusing Claude, including those developing software for conventional weapons like firearms, missiles, drones, bombs, and control systems in China, Russia, and Yemen. The company detected and neutralized activity linked to affiliates of the ShinyHunters cybercrime collective, a prolific enterprise involved in attacks on major corporations globally.
Jacob Klein, Anthropic's threat intelligence head, noted that AI models have become more potent over the past year, increasing the associated risks.
Written by urgent.news from Free Malaysia Today's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
- Anthropic disrupts Russian, Chinese AI campaigns targeting Claude models freemalaysiatoday.com
