Anthropic has reported several cases in which its Claude AI models were allegedly misused for weapons development, cyber operations and intelligence-related activities. The company said actors in Yemen, China and Russia used Claude for work connected to conventional weapons, including missiles and armed drones.Anthropic also identified a suspected Russia-linked cyber espionage campaign targeting Ukrainian government, military and diplomatic officials. The findings come from its latest threat intelligence report covering malicious activity over the past eight months. The report highlights growing concerns about how advanced AI can support cybercrime, espionage and other high-risk activities.How Claude Was Misused GloballyAnthropic said the latest cases show that AI misuse is moving beyond simple chatbot questions. According to the company, some operators used Claude in larger workflows involving software development, intelligence gathering and cyberattacks. The reported activity involved actors connected to Yemen, China and Russia, while Anthropic said it disrupted the identified misuse and strengthened its safeguards.Claude was reportedly used in software development linked to conventional weapons.Anthropic identified activity involving missile and armed-drone systems.A suspected Russia-linked group used AI during cyber operations against Ukraine.Chinese AI labs were accused of attempts to extract Claude’s capabilities.Anthropic said it banned accounts and improved its security controls after detection.Claude Misuse Raises Security ConcernsAnthropic’s report covers several types of misuse, including biological research, conventional weapons development, cyberattacks and intelligence operations. The company said operators in China, Russia and Yemen used Claude for weapons-related software or to support intelligence and procurement activities connected to weapons programmes.The report also described a suspected Russian-linked campaign against Ukrainian government, military and diplomatic targets. Anthropic said AI was used across several stages of the operation, including phishing and malware-related activity. The company linked the reported techniques to Midnight Blizzard, a threat actor previously associated by the US government with Russia’s SVR foreign intelligence service.Also Read: OpenAI Safety Warning: Paul Christiano Says AI Risks Need Stronger SafeguardsNew Report Details Global ThreatsThe latest findings are significant because Anthropic says malicious users are increasingly treating AI as an active tool rather than simply a source of advice. The company reported that some attacks involved multi-agent systems, allowing AI to handle larger portions of complex cyber operations.Anthropic also said it disrupted attempts by several China-based AI laboratories to extract Claude’s capabilities. One operation attributed to Alibaba reportedly involved more than 151 million exchanges between May and July 2026. China’s foreign ministry said it was not aware of the report and rejected what it described as distortion of facts about China.Key HighlightsAnthropic reported Claude misuse linked to weapons development in Yemen, China and Russia.Reported weapons-related activity included missiles, armed drones and other conventional weapons.A suspected Russia-linked operation targeted Ukrainian officials and institutions.AI was reportedly used across multiple stages of cyberattacks rather than only for basic assistance.Anthropic said it disrupted the identified activities and strengthened safeguards.The findings add to wider concerns about AI safety, cybersecurity and international espionage.AI Misuse Raises Global Security QuestionsAnthropic’s latest report shows how quickly advanced AI tools can become part of sophisticated cyber and weapons-related activity. The company says it is improving safeguards, monitoring and threat intelligence to detect such misuse. The findings also underline the challenge faced by the wider technology industry as AI becomes more capable.Disclaimer: This article is provided for knowledge purposes and to share news information.
