OpenAI discloses novel encryption bypass in distillation attack linked to MoonshotAI

OpenAI revealed a novel encryption bypass technique used in a model distillation attack, attributing parts of the activity to individuals associated with Chinese firm MoonshotAI without providing hard evidence. The attack targets AI model intellectual property, highlighting a new vector in AI security. The lack of public evidence and the attribution claim introduce uncertainty, but the disclosure signals escalating adversarial activity against AI systems.