Pangram’s Quest to Quantify AI in Writing Faces Trust and Bias Tests
WIRED profiles Pangram, a small Brooklyn AI-detection startup that scores how much a text is AI-generated using synthetic mirroring and hard negative mining. After high-profile AI-writing controversies and a Substack integration, Pangram markets itself as a trusted arbiter of originality across publishing and education, but critics warn of false positives and systemic biases against non-native writers, while the company stresses transparency and ongoing refinement.
