Cybersecurity2 min reading time
OpenAI says Moonshot-linked users tried to extract its AI reasoning
Covered by 3 sources
Read the full article
OpenAI identified a coordinated campaign linked to Moonshot AI attempting to extract its models' hidden reasoning via adversarial distillation, involving thousands of users and tens of thousands of requests in July 2026. OpenAI blocked the activity, enhanced protections, and shared findings with other AI labs and government bodies.
