Pangram Raises $9M to Combat AI-Generated Content Flooding the Internet

Summary

AI detection startup Pangram has secured $9 million in Series A funding led by Menlo Ventures to expand its tools for identifying AI-generated text and images. The New York-based company, founded by Stanford AI and machine learning graduates Max Spero and Bradley Emi roughly two years ago, is simultaneously launching Pangram 4, a next-generation text detection model claiming over 99% accuracy at identifying AI-assisted writing and mixed human-AI content, along with Pangram Image for image detection currently in research preview.

Pangram's detection technology trains large machine learning models on tens of millions of known human documents, then creates synthetic "mirrors" of each using frontier LLMs to learn the distinctive stylistic choices that differentiate AI-generated content. Unlike approaches relying on metadata or watermarks, the model learns high-confidence patterns in how AI consistently writes. Crucially, Pangram goes beyond binary detection to identify varying levels of AI assistance—such as humans using AI tools for editing—which founder Spero believes should be disclosed to readers.

The funding round includes participation from Haystack, ScOp, Script Capital, and Cadenza. The raise arrives as institutions increasingly enforce policies around undisclosed AI use: arXiv banned authors for one year if they fail to properly review LLM outputs, while courts have begun sanctioning lawyers who cite AI-hallucinated sources. Pangram faces competition from GPTZero, Originality.ai, Copyleaks, and Winston AI, but the market for content verification tools is expanding rapidly as AI-generated content pervades search results, social media, and academic institutions.

• Institutional adoption accelerating: arXiv, courts, and platforms implementing policies that create growing demand for reliable AI detection tools