As AI Content Floods the Internet, Pangram Raises $9M to Detect It

As artificial intelligence floods the internet with machine-generated text, New York-based startup Pangram is stepping up to fight the "AI slop infestation" spreading across the web. The company has raised $9 million in a funding round led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital, and Cadenza. The investment comes as demand for tools that reliably distinguish human-written content from AI-generated text continues to grow — a trend expected to intensify through 2026 and beyond. ## A New Generation of AI Detection Alongside the funding announcement, Pangram is launching its next-generation AI text detection model, Pangram 4, and debuted an AI image detection model called Pangram Image. The new text detector boasts over 99% accuracy in identifying AI-assisted writing and mixed human-AI content. It can also more easily detect AI "humanizer" programs — tools that attempt to disguise machine-generated text as human-written. The AI image detector is currently available only via research preview, with a wider release planned in the coming weeks. ## Founded in the Wake of ChatGPT Stanford AI and machine learning graduates Max Spero and Bradley Emi founded Pangram about two years ago, shortly after OpenAI launched ChatGPT. The chatbot opened the floodgates for an internet increasingly dominated by bots, generic AI-generated SEO content, and what Spero describes as "LLM-powered Russian disinformation campaigns and UAE-influenced campaigns on Twitter." "I think it’s just incredibly valuable to know whether what you’re looking at is something that’s AI-generated or not," Spero told TechCrunch. "Especially text that you’re reading, because it changes how people approach the text. Is this something I’ll have to watch out for hallucinations and jump in skeptically, or is this something that I trust was well-researched from an actual journalist?" ## How the Detection System Works Pangram’s AI detection system is built on a large machine learning model trained on tens of millions of known human-written documents. For each document, the startup created a "synthetic mirror" — a version that replicates the topic, length, and tone of voice but is written by a frontier LLM. The model learns the stylistic differences and consistent choices that AI makes, allowing it to identify AI-generated content with high confidence. Crucially, Spero notes that the detector does not rely on copy-paste metadata or hidden watermarks. Instead, it analyzes the writing itself for subtle patterns that distinguish AI from human prose. ## Beyond Binary Detection For Pangram, AI detection isn’t just about determining whether a piece of text was entirely written by a machine. The system also distinguishes between levels of AI assistance — for example, when someone writes their own content but then asks an AI to edit or polish it. Spero believes that AI assistance can be acceptable, provided the writer discloses its use. This nuanced approach could become increasingly important as more professionals integrate AI tools into their workflows while maintaining transparency. As 2026 approaches, with AI-generated content expected to make up an even larger share of online material, Pangram’s technology — and its fresh $9 million — positions the startup to play a key role in restoring trust in digital information.

via TechCrunch