San francisco: Anthropic has reported the shutdown of numerous state-sponsored surveillance operations employing artificial intelligence (AI) models, with a focus on countries such as China, Iran, and regions in West Africa. These operations, which targeted ethnic minorities and dissidents, were identified and disrupted between January and July.According to France24.com, the surveillance campaigns specifically aimed at pro-democracy figures in Hong Kong, Tibetan and Falun Gong communities across Asia, and Iranian minority communities and regime opponents abroad. In one instance, Iranian actors devised methods to identify individuals via social media, while a Malian contractor used Claude, Anthropic's AI model, to develop intelligence-gathering software.The report highlighted the substitution of AI for traditional engineering tasks in conducting these surveillance activities. In addition to espionage, Anthropic intervened in attempts to use AI for weapon design, questionable biological research, and dating scams.The AI lab also accused Chinese developers of using Claude to generate responses while covertly extracting the data to enhance their own AI models. This practice, known as "distilling," has previously been linked to unauthorized resource usage from American labs like Anthropic and OpenAI.Anthropic cited cases involving Chinese firms Moonshot and Deepseek, who allegedly used Claude to process a substantial volume of customer requests through fraudulent accounts, with some data potentially breaching privacy agreements.The report also mentioned Anthropic's intervention in research related to biological weapons, including work on viruses and toxins. However, the AI lab refrained from disclosing the identities of the scientists involved, emphasizing a lack of intent to cause harm.