Pangram Labs, based in Brooklyn, develops machine learning classifiers for AI content detection. Its Pangram 4 text model, released in July 2026, is a transformer classifier that returns both document-level and segment-level AI versus human fractions across more than 20 languages, covering output from ChatGPT, Claude, Gemini, Llama and Grok. An accompanying image detector shipped as a research preview covering GPT Image, Gemini Nano Banana, Midjourney, FLUX and Grok Imagine, plus video from Kling AI and Veo. The product is delivered as a web app, a Chrome extension that scans social feeds in place, LMS integrations for Google Classroom, Canvas, Brightspace and Moodle, and a documented REST API with an open-source Python SDK. Customers span academic integrity, editorial vetting, platform moderation and recruiting; a University of Maryland study of 186,000 articles from 1,500 US newspapers used Pangram to estimate AI-generated content prevalence. Accuracy figures published by the company are internal benchmarks.
Funding
Pangram raised a $1.25M pre-seed led by Haystack VC and a $2.7M seed led by ScOp Venture Capital, both announced together in June 2025. In July 2026 it raised a further $9M led by Menlo Ventures with participation from Haystack, ScOp, Script Capital and Cadenza, bringing total funding to roughly $13M. The 2026 round was not given a public series designation.
