From Left to Right: The Midhun Danmudi, Edward Tian and Robert Nicolaides. Aaron mok for observer
Tucked away inside a WeWork meeting room in Lower Manhattan, Edward Tian pulls up a Google Doc on his laptop titled “Secret Master Plan,” a draft outlining his startup’s growth strategy. The 25-year-old co-founder and CEO of GPTZero taps the trackpad and grins. “Eighty-three percent is me. Seventeen percent is my co-founder Alex,” Tian said, nodding to the authorship breakdown his software just spit out.
The forensic authorship analysis provides a glimpse into what GPTZero is becoming. The tool that first went viral for helping teachers catch homework is now pivoting into the opposite side of the game: helping users write better and more responsibly with A.I. “We want to make GPTZero the de-facto tool for teachers—writing professors, science professors, essay instructors—to help students learn how to write and cite academic work,” Tian told Observer.
“We’re not anti-A.I. We’re A.I.-forward,” Tian emphasized. That shift in strategy comes as generative A.I. encroaches on every part of life, sometimes leading to unintended consequences. One MIT study recently found that students who relied on A.I. to write showed lower brain activity in areas linked to memory and critical thinking. Schools across the U.S. are cracking down on A.I.-generated assignments. Online, LinkedIn posts that call out A.I.-written content for sounding stiff or soulless keep racking up views.
Beyond flagging A.I.-generated text, the platform now offers real-time suggestions on clarity and structure, checks citations for accuracy, and helps teachers figure out who contributed what in group projects. It’s less about catching cheaters, according to Tian, and more of a “pedagogical tool” to hone writing and critical thinking skills. As generative A.I. becomes harder to avoid, GPTZero’s challenge is to help people work with the technology in ways that sharpen their thinking—without letting machines replace it.
From college thesis to startup
Raised by software engineers between Beijing and Toronto, Tian’s passion for technology and learning was instilled in him at a young age. GPTZero began as Tian’s senior thesis at Princeton University, where he majored in computer science and minored in journalism. After getting early access to OpenAI’s language models during a Microsoft internship, his interests shifted. A month after ChatGPT launched in late 2022, Tian quickly built GPTZero and tweeted it out. Within hours, it went viral.
When GPTZero first launched, many doubted whether it could survive. Skeptics questioned the longevity of the company, writing off the launch as a mere “project” over a serious business venture. That spooked initial investors. But over the years, the tool has grown far beyond those early expectations. In 2024, users scanned more than 250 million documents, according to Tian. By mid-2025, that number had more than doubled. Today, GPTZero generates $20 million in annual recurring revenue in paid subscription plans and partnerships.
The company has now grown to a 23-person team between New York and Toronto. The Canadian office location was chosen, in part, because of its strong machine learning community and access to top tier talent from neighboring universities, Tian said.
The software maintains between 100,000 and 200,000 daily active users and has seen over a million daily users during peak exam seasons, according to Tian. It’s not just students and teachers using the tool. Marketers, grant reviewers and even hiring managers are also adopting the tool to improve their writing. “We’re still constantly being underestimated,” Tian said.
Educators remain a key user base, and they’re warming up to using GPTZero as more than just an A.I.-detection tool. Earlier this year, GPTZero launched a six-part certificate webinar series called “Teaching Responsibly with A.I.,” drawing over 3,000 teachers eager to learn how to thoughtfully integrate A.I. into their classrooms. With a 97 percent satisfaction rate, the webinars led to more than 500 teachers adding GPTZero to their syllabi, instructing students on how to use its tools to improve their writing and cite sources accurately, according to Tian.
Now, students using GPTZero’s Docs Chrome extension to receive feedback on their writing, such as flagging vague ideas or suggesting clearer phrasing. At the same time, teachers use features like typing pattern analysis to track contributions in group work and a bibliography checking tool to flag missing or made-up citations.
The emphasis on A.I. as an instructional tool has resonated with some teachers. Tian recounted a conversation with one teacher who said they used to pay close attention to grammar when grading. Now, grammar isn’t their main concern because grammatical mistakes actually may indicate the students didn’t use A.I. With GPTZero, the teacher could focus on what truly matters: the quality of writing. “That’s a great equalizer,” Tian said.
A trust dilemma
GPTZero’s expansion comes at a moment when businesses and universities are still figuring out how to adopt A.I. Some are eager to embrace it for the productivity gains; others are more cautious, unsure of where to set up guardrails. That tension is especially present in schools, where educators are navigating murky territory without much guidance.
Without clear norms around responsible use, some students may be discouraged from experimenting with A.I. altogether. And despite GPTZero’s shift toward a more constructive role, its roots in plagiarism detection raises concerns that its adoption could reinforce a culture of surveillance and mistrust.
In response to these concerns, Tian emphasizes that GPTZero is moving away from a “plagiarism mentality” toward promoting mindful A.I. usage.
That philosophy also shapes the way the team works. Before joining GPTZero, head of design Robert Nicolaides was hesitant to use A.I. in his creative process. Now, he’s less skeptical. He uses chatbots for tasks like transcribing user research and generating “dummy text” for mockups. However, he finds A.I. less useful for brainstorming original design ideas.
Experimenting with A.I. to see where it helps and where it falls short is the same approach many teachers take, Nicolaides said. “What we’re doing here is about responsible A.I. use and helping people improve their writing—not do the work for them,” he told Observer.
Full-stack engineer Midhun Dandamudi said A.I. has “multiplied” his productivity. Even though he’s aware of the harm of overreliance, he believes that A.I. is here to stay, and users must find a way to embrace it in a critical way. “If you build the right tools, you can help people use them responsibly,” Dandamudi told Observer.
Looking ahead, GPTZero is developing features to surface credible sources, reduce hallucinated content, and help teachers integrate A.I. into lesson plans, said Tian. While writing remains the focus, some users are already asking for tools to detect A.I.-generated code because of concerns that the code may not work or be vulnerable to attacks.
Responsible A.I. use will remain central to GPTZero’s design. But when it comes to deciding how A.I. fits into the classroom, Tian says that’s ultimately up to teachers to decide. GPTZero’s role, as he sees it, is to give them what they need while encouraging them to think broadly about how to use the technology in ways that help—without causing harm.
“We’re in a pretty unique position to do that,” Tian said.
