A startup just raised $9M to fight AI text — its logic is simple: humans aren't multiplying, GPUs are

New York startup Pangram raised a $9 million Series A led by Menlo Ventures — the company builds software that tells AI-generated text and images apart from human-made ones, TechCrunch reports. The round also drew Haystack, ScOp, Script Capital, and Cadenza.

The company was founded roughly two years ago by Stanford AI/ML graduates Max Spero and Bradley Emi. Its flagship model, Pangram 4, detects fully AI-written text, human-AI hybrid writing, and attempts to "humanize" machine text, all with over 99% accuracy. There's also an AI image detector in research preview, a browser extension that checks posts on X, LinkedIn, Substack, Reddit, and Medium, an API for enterprise clients, and a $20-a-month web subscription.

“We're getting new GPUs faster than new people are being born. If we do not actively discriminate in favor of human content, then we're just gonna see more and more AI”

— Max Spero, Pangram co-founder

• The model is trained on millions of real documents, each paired with a synthetic counterpart, so the system learns the stylistic gaps between the two
• Clients include Substack, Quora, schools, universities, publishers, and recruiters
• In the AI-content-detection market, Pangram competes with Winston AI, Originality.ai, Copyleaks, and GPTZero

Demand for tools like this is growing alongside the problem itself: telling an AI photo from a real one, or AI text from human writing, keeps getting harder — we've covered the specific signs that still work here.

This piece is informational, not a recommendation to buy, sell, or hold any asset.