OpenAI Releases Annual Report on Malicious Uses of AI in June 2025

OpenAI has published its annual report on the malicious uses of AI in June 2025. By using AI as a force multiplier for its investigation teams, OpenAI has detected, disrupted, and exposed abusive activities such as social engineering, cyber espionage, deceptive employment schemes, clandestine influence operations, and scams. Four out of the ten reported cases appear to originate from China, involving social engineering, clandestine influence operations, and cyber threats. Other abuses were detected in various countries, including a task scam likely originating from Cambodia, comment spam apparently from the Philippines, and clandestine influence attempts potentially linked to Russia and Iran. The report highlights that the AI models used by malicious actors could evolve rapidly, reducing future visibility on these threats.