OpenAI said it banned accounts linked to Chinese law enforcement, romance scammers, and influence operations, including a smear campaign against Japan’s first woman prime minister, in a report detailing the misuse of its ChatGPT technology.

The company said several accounts used its chatbot alongside other tools, including social media accounts, to carry out cybercrimes while posing as a dating agency, law firms, and U.S. officials, among others.

Here are some details from OpenAI:

• A small set of accounts that likely originated in China used OpenAI’s models to request information about U.S. persons, online forums, and federal building locations, and sought guidance on face-swapping software.
• The same accounts also generated English-language emails to state-level U.S. officials or policy analysts working in business and finance, inviting targets to participate in paid consultations.
• OpenAI said it banned a ChatGPT account linked to an individual associated with Chinese law enforcement whose activity involved orchestrating a covert influence operation targeting Japanese Prime Minister Sanae Takaichi.
• A cluster of ChatGPT accounts used the chatbot to run a dating scam targeting Indonesian men and likely defrauded hundreds of victims a month, according to OpenAI.
• OpenAI said the scam used ChatGPT to generate promotional text and ads for a fake dating service, luring users to join the platform and pressuring targets to complete several tasks requiring large payments.
• Several accounts used OpenAI’s models to pose as law firms and impersonate real attorneys and U.S. law enforcement, targeting fraud victims, OpenAI said.

(Reporting by Deborah Sophia in Bengaluru; Editing by Tasim Zahid)