Anthropic Blocks Russian Hackers, Chinese AI Labs Accused of Misusing Claude
- Anthropic disrupted various malicious uses of its Claude AI models over eight months, including attacks linked to Russia and efforts by Chinese AI firms to replicate Claude's capabilities.
- Seven China-based labs, including Alibaba and Xiaomi, were accused of illicit activities involving Claude, such as routing live customer chats through Claude for training data.
- A Russia-linked group named Midnight Blizzard reportedly used AI extensively in cyber espionage against Ukrainian government and military targets, employing complex malware evasion techniques.
- Anthropic found Claude being misused to develop software for conventional weapons and control systems in countries such as China, Russia, and Yemen.
27 Articles
27 Articles
China's I.E. DeepSeek and Moonshot re-routed their users' requests to neuralize Claude instead of processing them using their own models, as described in the report of Anthropic (Claude Developer).
The essence of the process, called distillation, is for the rival company to uncover the chatbot's thought process and develop its own AI based on it.
According to the developer Anthropic, artificial intelligence is increasingly being used to control entire cyberattacks. The company reports of a suspected Russian spy campaign and Chinese AI companies who have abused its model for their own purposes.
Anthropic Says Russian, Chinese Threat Actors Used Its AI Model Claude for Malicious Activity
Claude is a series of large language models developed by Anthropic. It is also the name of the chatbot that uses these models to interact with users. Above, an illustration of Claude on June 18, 2026. (Riccardo Milani/ Hans Lucas/AFP via Getty Images)
Actors who use Claude to develop weapons software and Chinese technology giants who want to improve their own AI models: Anthropic wants to prevent this misuse.
Cheap AI from China raises doubts about the market opportunities of major US providers such as Anthropic. The company now reports about bold tricks by Chinese competitors. Sensitive data came from Russia to the USA.
Coverage Details
Bias Distribution
- 39% of the sources lean Left, 38% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
