New York-based AI detection startup Pangram has the mission of combating AI Debris Infestation is spreading across the Internet and just raised $9 million in a bet that demand for tools that distinguish human-generated content from AI-generated text will only grow.
Pangram’s fundraising, led by Menlo Ventures, with participation from Haystack, ScOp, Script Capital and Cadenza, comes as the startup also launches its next-generation AI text detection model, Pangram 4, and an AI image detection model, Pangram Image.
Pangram says the new text detection model is over 99% accurate at finding AI-assisted writing and mixed human-AI content, and can more easily detect humanizing AI programs. The AI Image Detector is only available through the Research Preview for now; Pangram plans to roll it out more widely in the coming weeks.
Max Spero and Bradley Emi, Stanford AI and machine learning graduates, launched Pangram about two years ago, after the launch of ChatGPT opened the floodgates for an Internet filled with bots, AI-generated SEO content, and what Spero calls “LLM-fueled Russian disinformation campaigns and UAE-influenced Twitter campaigns.”
“I think it’s incredibly valuable to know whether what you’re seeing is something AI-generated or not,” Spero told TechCrunch. “Especially the text you’re reading, because it changes the way people approach the text. Is this something I’ll have to watch for hallucinations and jump at with skepticism, or is this something I trust was well researched by a real journalist?”
Pangram’s AI detection system is essentially a large machine learning model that was trained on tens of millions of known human documents. The startup then created a “synthetic mirror” for each document, replicating the topic, length and tone of voice, but written by a cutting-edge LLM.
“Our model is learning the stylistic differences and choices that AI consistently makes and is able to use that to learn what makes something AI-generated with high confidence,” Spero said, adding that the AI detector doesn’t rely on copy-and-paste metadata or hidden watermarks.
For Pangram, AI detection is not just about whether a piece of text was written entirely by AI. It’s also about distinguishing between levels of AI assistance, such as someone writing something themselves but then asking AI to edit or clean it up. Spero believes that AI assistance may be acceptable, as long as the writer discloses their use of AI.
The appearance of Pangram comes at a time when the use of AI is becoming more common. In some cases, such as Canadian politician who read an AI message Loudly in a speech to legislators, mistakes result in ridicule. In other cases, such as certain lawyers who present their cases using fake quotes created by ChatGPT, the consequences could be sanctions and fines.
That reaction is not only costing individuals shame or sanctions: it is also beginning to appear in institutional rules.
The open-access archive arXiv introduced a new compliance policy this year, stating that submissions containing evidence that the authors did not review the LLM results (such as hallucinated references or meta-comments like “Do you want me to make any changes?”) can trigger a one-year submission ban.
Pangram isn’t the only one betting that AI detection will become more sought after. Competitors like Winston AI, Originality.ai, Copyleaks and GPTZero are pursuing the same demand and each builds its own detector.
Pangram’s technology, while not perfect, could help fuel resistance to accepting the AI-generated content flooding the Internet, the courts, and academic articles.
Users can access Pangram through a $20/month web subscription or download the Chrome extension, which automatically tags posts in real time on X, LinkedIn, Substack, Reddit, and Medium. It also provides a feed health score with a percentage breakdown of human versus AI content on your screen.
Pangram also offers its technology via API. Notably, Substack recently integrated Pangram technology on its platform to show readers which of their favorite authors write their newsletters using AI. Other API customers include Quora, schools and universities, publishers, agents and recruiters, among others, according to Spero.
Does Pangram work?
Spero said that about one in 10,000 human documents is incorrectly labeled as AI using Pangram’s model, so I decided to put it to the test. The text detection model was very impressive but not perfect. It easily flagged fully AI-generated news articles written by both ChatGPT and Claude, and was rarely fooled by my attempts to edit AI-generated text to make it sound more human. At the same time, Pangram flagged sentences I completely rewrote as written by AI. Pangram was also not at all fooled by my attempts to encourage ChatGPT and Claude to evade AI detectors when generating content.
I also gave ChatGPT and Claude one of my own articles and asked them to polish it. Pangram gave it an AI-assisted score of 13%, which was probably close to accurate, but the model was able to detect subtle changes in word choice in some sentences and ignored them in others. It also marked some sentences as AI-assisted when they were written by humans. That was notable because when I gave Pangram the same article in its entirety, just as I had written it, it got a 100% human score.
Perhaps the problem was that news articles can be a bit boring and can easily look like AI. So I tried a different tactic. I tried Pangram on my own, with the more sonorous and personal content from the Substack newsletter, pasting the first half of the text into Pangram and then asking ChatGPT and Claude to copy my style and write the second half. For the most part, Pangram easily detected human-written text versus AI-written text.
My limited testing of Pangram’s new image detection model turned out to be equally impressive.
Pangram’s AI image detection system promises to detect AI-generated images in AI models, unlike watermark-based checks from OpenAI or Google DeepMind, which mostly detect their own output. It works with pixel-level distributions and learns subtle statistical differences between real photographs and AI-generated images. Spero says the model can even detect an AI image that appears inside a real-world photograph.
In my tests, the model easily detected AI-generated images, whether photorealistic or cartoonish. I can also confirm that the model was able to detect an AI image appearing in a real-world photo (the heatmap Pangram provides clearly illuminates the image), although in one case it incorrectly labeled a photo of an AI-generated image as human content.
Spero says he doesn’t want his technology to fuel a witch hunt against people who use AI to write, but there needs to be some kind of mechanism to counteract this.
“The future I see is that AI content will continue to proliferate,” Spero said. “We are getting new GPUs faster than new people are being born. If we don’t actively discriminate in favor of human content, then we will see more and more AI, and it will drown out any human signals we have.”
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
