State-backed hackers are experimenting with OpenAI models 

Microsoft and OpenAI report that state-backed hacking groups from China, Iran, North Korea, and Russia are experimenting with large language models for tasks including technical reconnaissance (e.g., satellite and radar research), generating spearphishing content, and assisting with script/code development and troubleshooting. The companies observed these behaviors as early-stage, productivity-oriented uses rather than evidence of significant, LLM-driven attacks on the monitored platforms, while warning that activity on less-monitored or open-source models may not be captured.