{"slug": "pangram-raises-9m-as-substack-bets-readers-will-pay-for-human-writing", "title": "Pangram raises $9M as Substack bets readers will pay for human writing", "summary": "Pangram, a New York-based AI detection startup, has raised $9 million in a funding round led by Menlo Ventures, bringing its total funding to roughly $14 million since its founding in 2023. The company's detection tool, now integrated into Substack's platform to scan posts for AI involvement, claims a 0.01% false positive rate verified by researchers at the University of Chicago and the University of Maryland. The investment signals growing venture conviction in detection-as-a-service as AI-generated content becomes indistinguishable from human writing.", "body_md": "*With Menlo Ventures backing Pangram's latest $9M round and Substack rolling out AI-content scanning on July 21, the business of telling humans from machines is becoming infrastructure.*\n\nWhen Substack launched its AI detection feature this month, co-founder and CEO Chris Best had a name ready for what he was fighting: \"Claudefishing.\" His term for the gap between a reader's assumption that a human wrote something and the reality that no one did. The tool powering it was Pangram, a New York-based AI detection startup that has now raised $9 million in a round led by Menlo Ventures, bringing its total funding to roughly $14 million since it was founded in 2023.\n\nThe Substack integration is available on web and iOS, with Android support promised shortly. Readers can request a scan of any post or note longer than 100 words, and Pangram returns an estimate of AI involvement. Pangram said its detection model achieves a 0.01% false positive rate, a figure the company says has been independently verified by researchers at the University of Chicago and the University of Maryland. Best was more careful about what the tool actually promises: it can estimate whether AI was used, he noted, not whether a post reflects genuine human thought or care.\n\nThe funding lands alongside two significant product releases. Pangram 4, an updated text-detection model, is now live. The company has also released an AI image detection model in research preview, claiming 99.5% accuracy - a significant expansion beyond text at a moment when synthetic images are flooding social platforms and news feeds. For investors, the pitch is straightforward: as AI-generated content becomes indistinguishable to the naked eye, detection becomes infrastructure, not a feature.\n\n## From pre-seed to conviction capital\n\nPangram's earlier rounds tell the trajectory. A $1.25 million pre-seed was followed by a $4 million seed in June 2025, led by Haystack and ScOp Venture Capital. Then a further $2.7 million, focused specifically on scaling in edtech and media. Pangram also partnered with Qwoted, a journalist sourcing platform, to help reporters spot AI-generated pitches - a problem that has become acute in PR. None of those rounds were splashy. The $9 million from Menlo is the first that signals the category has moved past experiment into something VCs are willing to stake real conviction on.\n\nThe comparison to bot detection is useful here. Spur, which tracks bot networks and coordinated inauthentic behavior, built a durable business long before most platforms took the problem seriously. Detection-as-a-service tends to look niche until the thing it detects becomes a genuine crisis for platforms that depend on trust. Substack's entire economic model rests on the idea that readers will pay recurring subscriptions for content written by specific humans. Claudefishing is not a philosophical concern for them. It's an existential one.\n\nThat said, no AI detector is reliable enough to convict. Lightly edited or paraphrased AI text is far harder to catch than raw model output, and the history of detection tools in adjacent spaces - plagiarism detection, bot scoring - is littered with false positives that damaged real people. Pangram's 0.01% false positive claim is striking, but even that rate, applied at Substack's scale, produces a non-trivial number of wrongly flagged human writers. Best acknowledged this by framing the scan as an estimate rather than a verdict, which is the honest position and also a useful legal one.\n\n## Edtech, media, and who controls the verification layer\n\nEdtech is the other engine behind Pangram's growth. Colleges and secondary schools have been fighting AI-assisted submissions since ChatGPT launched in late 2022, and most of the early detection tools failed badly - with GPTZero and others generating academic witch-hunts before retreating to more qualified probability language. Pangram's pitch to edtech buyers is the same as its pitch to media: high accuracy, low false positive rate, independent verification. Whether the classrooms where it's already deployed bear that out is harder to know from the outside.\n\nWhat's clear is that detection has become its own category. Publishers need it to protect subscription trust. Edtech companies need it to maintain academic integrity policies. Media organisations need it to verify what's in their intake pipelines. Pangram is betting it can be the infrastructure layer under all three, with Menlo Ventures apparently persuaded that the market is big enough to justify a lead position. The arms race framing is accurate but incomplete. This isn't just about catching AI slop. It's about whether authenticity itself becomes a verified, sellable attribute in the information economy - and who controls the verification layer.\n\n**Also read:** [Google DeepMind disbanded its Nobel Prize-winning AlphaFold team and the talent went straight to Anthropic](https://startupfortune.com/google-deepmind-disbanded-its-nobel-prize-winning-alphafold-team-and-the-talent-went-straight-to-anthropic/) • [OpenAI's own AI agents broke out and hacked Hugging Face and a second company](https://startupfortune.com/openais-own-ai-agents-broke-out-and-hacked-hugging-face-and-a-second-company/) • [Dwelly raises $170M to buy up UK letting agencies one by one and run them on AI](https://startupfortune.com/dwelly-raises-170m-to-buy-up-uk-letting-agencies-one-by-one-and-run-them-on-ai/)", "url": "https://wpnews.pro/news/pangram-raises-9m-as-substack-bets-readers-will-pay-for-human-writing", "canonical_source": "https://startupfortune.com/pangram-raises-9m-as-substack-bets-readers-will-pay-for-human-writing/", "published_at": "2026-07-29 11:58:30+00:00", "updated_at": "2026-07-29 12:20:11.610848+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-products", "ai-startups", "ai-tools", "ai-ethics"], "entities": ["Pangram", "Menlo Ventures", "Substack", "Chris Best", "University of Chicago", "University of Maryland", "Haystack", "ScOp Venture Capital"], "alternates": {"html": "https://wpnews.pro/news/pangram-raises-9m-as-substack-bets-readers-will-pay-for-human-writing", "markdown": "https://wpnews.pro/news/pangram-raises-9m-as-substack-bets-readers-will-pay-for-human-writing.md", "text": "https://wpnews.pro/news/pangram-raises-9m-as-substack-bets-readers-will-pay-for-human-writing.txt", "jsonld": "https://wpnews.pro/news/pangram-raises-9m-as-substack-bets-readers-will-pay-for-human-writing.jsonld"}}