Tag: AI governance
-
Analysis: What the DeepSeek Audit Can—and Cannot—Tell Us About AI Information Control
Research analysis separating training data, retrieval, moderation layers, routing and final output when interpreting politically sensitive DeepSeek behavior.
-
News: New DeepSeek Audit Documents Content-Sensitive Response Differences in Controlled Tests
New Free China Movement research documents controlled black-box DeepSeek tests involving Tiananmen Mothers, the 1989 Beijing student movement and Lian Shengde, while explicitly limiting conclusions to observed behavior.