Google DeepMind and University of Pennsylvania Test AI Text Detector

2 outlets reporting

Your call?🐂 Bullish🐻 Bearish

75.4 F1

detector score

FT-Model+SAE scored 75.4 F1 on an out-of-distribution COLING test split, compared with 75.2 for a fine-tuned Gemma-2-2B baseline. The system pairs model features with human-readable descriptions.