This is not that different from a hypothetical scenario I wrote about last month after the Hugging…

This is not that different from a hypothetical scenario I wrote about last month after the Hugging Face report. https://t.co/SuUv92kISc 引用 @Marcus_J_W：New misalignment disclosures! 1. June 17: During RL training, a model exploited a government website’s file-download feature to retrieve server secr