Superhuman Acquires GPTZero, Putting AI Detection in the Inbox
Edward Tian built GPTZero in early 2023 as a Princeton undergraduate worried that teachers could no longer tell whether their students had written their essays. Three years later, the detection tool he coded over a winter break has 19 million registered users, and it has been acquired by Superhuman, the email company, in a deal whose terms were not disclosed.
Superhuman announced the acquisition on Tuesday, saying GPTZero’s technology will be woven into its email platform to flag messages that appear to be generated by artificial intelligence and to score the authenticity of incoming mail. The deal pairs a startup that made its name exposing AI-written text with an email client that is itself an AI product, part of the broad industry scramble to decide what role detection will play in a world flooded with machine-generated content.
GPTZero’s rise was one of the fastest in consumer software. What began as a thesis project became a campus tool within months, then a global service used by teachers, admissions offices and journalists. The company raised $13.5 million from investors including Uncork Capital and Footwork, reached profitability in 2024 and built annual recurring revenue of about $30 million, according to figures the company has cited.
The business had a problem, though: growth slowed. The campus market that made GPTZero famous is finite, and its tools—which scan documents and flag statistical patterns typical of large language models—faced mounting skepticism as detection accuracy became a contested science. Studies showed that detectors misclassify non-native English writing and can be fooled by simple rewrites, and several academic institutions stopped relying on them entirely.
Superhuman, for its part, already had detection capability. The company took its current form last year when Grammarly, the writing-assistance firm, acquired the Superhuman email client and rebranded the combined company. Grammarly’s own AI-detection tools had been built for documents; Superhuman’s executives said the acquisition of GPTZero gives them a detection engine designed for the adversarial problem of the inbox, where spam, phishing and automated outreach increasingly arrive as fluent, AI-generated prose.
“Two detectors are better than one,” a Superhuman executive said, describing the deal’s logic. The company plans to let users see a probability score for whether an incoming message was machine-written, and to route suspected AI-generated bulk mail into a separate folder. Longer term, people familiar with the plans said, Superhuman wants to use GPTZero’s authorship analysis to help users understand who is actually writing to them—a real human, a bot, or a human running an AI assistant.
The acquisition is the latest in a wave of consolidation in AI-adjacent software, and it reflects a shift in how detection is being sold. GPTZero began as a product for catching cheaters. Its future, under Superhuman, is as infrastructure: a background check on every message, embedded in the tool where people read mail.
Tian, who will stay with the company, has described the mission in broader terms than plagiarism checking. He has argued that as AI-generated content multiplies, provenance—knowing what was written by a person and what by a machine—becomes a basic expectation of digital life. Superhuman’s platform, with its focus on professional users who live in their inboxes, gives that argument a commercial home.
The deal also illustrates the strange economics of the AI boom. GPTZero raised a fraction of what similar companies raised in the same period, yet it survived the collapse of interest in detection tools that followed the initial panic over ChatGPT. Its new owner faces the same fundamental question: whether people will actually pay for detection, or whether the market for authenticity turns out to be smaller than the panic suggested.
Superhuman executives said they believe the inbox is the right place to test that question. Email is where the consequences of undetected AI content are most concrete—fraud, phishing and manipulation—and where users already pay for software. GPTZero’s 19 million users, most of them teachers and students, give Superhuman a beachhead in education as well.
Analysts said the deal’s value likely reflects GPTZero’s stalled growth rather than its fame. One person familiar with the transaction said the price was “nowhere near” the valuations once floated for detection startups during the 2023 boom, when GPTZero’s early success made it one of the most-watched consumer AI companies in the world.
For Tian, the acquisition closes a loop. The tool he built to help teachers spot machine-written essays will now sit inside the product used to write and read professional correspondence, doing the opposite of what its original critics feared: instead of catching students, it will try to keep machines honest in the workplace. Whether that market is big enough to matter is the bet Superhuman just made.
