The article reports that Anthropic, the company behind Claude AI, uncovered a sophisticated global disinformation operation in which unknown threat actors used its chatbot to create and manage over 100 distinct fake political personas on Facebook and X. These AI-driven accounts engaged with tens of thousands of authentic users to promote politically-aligned narratives that supported or undermined interests in Europe, Iran, the United Arab Emirates, and Kenya. The operation used Claude not only for content creation but also as an orchestrator for engagement, deciding when bots should interact with real users. The campaign utilized advanced technology to mimic human behavior, scale its activities, and evade detection, and is suspected to be a commercial service for multiple clients. Anthropic also revealed other instances where its AI was abused for cybersecurity threats, recruitment scams, and malware development, warning that AI is making such campaigns more accessible and harder to counter.
Related articles:
Anthropic Tightens Rules to Curb Abuse of Claude AI
Cybercriminals Embrace AI to Launch More Sophisticated Malware Attacks
