Pangram Labs

Pangram Labs is a high-accuracy AI content detector that analyzes text segment by segment, attributes it to specific models, and now detects AI images.

Pangram is an AI detector focused on accuracy and reducing false positives, reporting figures around 99.7% accuracy on AI-generated samples and a false-positive rate near 1 in 10,000 in independent testing. Rather than a single verdict, it classifies each segment of a document into four tiers, from fully human to fully AI-generated, and attempts to identify which model likely produced flagged text. In 2026 it also introduced an image detection research preview. Independent research adds credibility: in June 2026, researchers at the University of Chicago Booth School of Business and Vrije Universiteit Brussel found accuracy between 99.8% and 100% on most models, and a peer-reviewed 2026 study rated Pangram the strongest overall performer against GPTZero, Turnitin, and Copyleaks. Importantly, reviewers note Pangram is strong on clean AI output but weaker once text is run through humanizers or blended with real writing, a limitation shared across the detection field. Pangram competes with detectors like GPTZero, Originality.ai, Turnitin, and Copyleaks. Its differentiation is high measured accuracy, segment-level analysis, and model attribution, though as with all AI detectors, no result should be treated as definitive proof, especially for consequential decisions like academic misconduct.