disrupting-malicious-uses-of-our-models-february-2025-update.pdf (7 MB)

OpenAI says it has disrupted malicious uses of its AI systems and is publishing case studies to support wider defensive work by governments, industry partners, and other stakeholders. The excerpt frames the threat set broadly, including authoritarian-state abuse, covert influence operations, scams, spam, child exploitation, and malicious cyber activity. It does not provide actor-specific evidence, malware details, infrastructure, or North Korea-specific case-study content in the available markdown excerpt. The usable finding is that AI-platform abuse investigations are being treated as a defensive capability against adversarial and authoritarian activity.