LLM Traces Exposed: AI Auditing, IP Face New Reality

The revelation of a method to extract "reasoning traces" from LLMs, published by AI researchers, fundamentally alters the landscape of AI auditing and intellectual property. This technique, which can identify the training lineage of models like those from China