Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
MetaPoisongithub.com · Low evidenceMetaPoison is a method for creating poisoned training data that causes neural networks to misclassify specific target images after training. The repository provides TensorFlow implementations for crafting such poisons and evaluating their effectiveness, including on real-world systems like Google Cloud AutoML Vision.
Parse Score