Anthropic said on September 10 that it has prevented many cases of using the company's artificial intelligence (AI) models to serve surveillance activities, and warned of the situation of taking advantage of AI to monitor minority communities and dissidents, AFP reported.
According to a report on the abuse of the company's model, incidents were detected and neutralized in the period from January to July 2026, originating from a number of areas including Iran and West Africa.
In one case, subjects in Iran developed a personal identification method through their social media accounts.
In another case, a contractor working for the Malian national security agency used Claude "to design a platform software to serve intelligence gathering activities".
AI is being used instead of a team of engineers," the report said.
Anthropic said that it has also prevented user attempts to design weapons, conduct suspicious biological studies and create dating scams.
