Pangram AI detection start-up raises $9 million and launches image tool
Pangram says its new model can flag AI text and images as publishers seek tools to filter automated content online.
By Marcus V. Thorne · Markets Editor
· 3 min read
Pangram AI detection start-up is raising $9 million as it releases Pangram 4.0, its latest text-detection model, and its first tool for identifying AI-made images. The New York company says demand is rising as publishers and other online operators try to distinguish human-created material from content produced by generative AI systems.
The financing is led by Menlo Ventures, with participation from ScOp Venture Capital, Haystack Ventures and other investors, according to Pangram. The round gives the company more capital at a time when AI-generated writing, posts and images are becoming a larger part of digital publishing and social media.
Pangram says its text-detection model can determine whether writing was generated by AI in 9,999 out of 10,000 cases. That claim, if sustained in commercial use, would put the company in a sensitive position for publishers, platforms and institutions that need to classify content but also face consequences when detection systems make errors.
What is Pangram AI detection?
Pangram builds software intended to identify whether text, and now images, were created by people or by AI systems. Such tools do not remove content by themselves; they provide a classification that customers can use when making editorial, moderation or verification decisions.
The company has signed high-profile online publishers as customers, according to The New York Times. Those clients are using Pangram in a market where generative AI has reduced the cost of producing large volumes of text and imagery, increasing the burden on organizations that rely on trust, attribution and quality control.
Max Spero, Pangram’s chief executive, has described himself as the internet’s “slop janitor,” a reference to low-quality or unwanted AI-generated material online, according to The New York Times. Spero co-founded the company with Bradley Emi. Both are 30 and met while studying as undergraduates at Stanford, where they worked in the university’s artificial intelligence lab.
After Stanford, Spero went to Google and Emi joined Tesla, where he worked under Andrej Karpathy, an AI researcher and OpenAI co-founder, according to The New York Times. The two later started Pangram after ChatGPT’s 2022 release, which accelerated public use of generative AI and expanded concerns about automated content appearing across the internet.
Why AI detection raises policy and business questions
AI-detection products sit at the intersection of publishing, education, platform governance and workplace communication. A tool that labels material as AI-generated can influence whether a post is trusted, whether an article is reviewed, or whether a creator’s work is treated as authentic.
Pangram’s own accuracy claim also highlights the risk. Even rare mistakes matter when detection results are used in decisions about writers, students, contributors or publishers. The company’s growth therefore reflects both a commercial opening and a broader governance problem: organizations want help identifying automated material, but they still need policies for how to act on probabilistic labels.
The new image-detection model expands Pangram beyond writing as synthetic media spreads across more formats. For publishers and platforms, that broadens the possible use cases from article screening to checks on visual material, though Pangram’s public announcement centers on the release of the product rather than detailed customer deployments for images.
This story draws on original reporting from NYT DealBook.
