- Stanford student Theo Baker argues in a New York Times guest essay that ChatGPT turned an existing culture of dishonesty at Stanford into the default norm.
- Stanford brought back proctored, handwritten in-person exams in spring 2026 — a practice banned for over a century — in response to rising plagiarism and cheating.
- Baker’s graduating class is the first to have spent its entire college career alongside ChatGPT, which launched roughly two months after Baker started in fall 2022.
- In a campus-wide survey, 49% of 849 Stanford CS majors said they’d rather cheat on an exam than fail.
What Happened
Theo Baker, a Stanford University student graduating in June 2026, describes in a guest essay for the New York Times how ChatGPT shaped his entire graduating class — turning an already existing culture of dishonesty into the default, The Decoder reported on Monday. Baker’s class is the first to have spent its entire undergraduate career alongside ChatGPT; the chatbot launched roughly two months after he started college in fall 2022.
Why It Matters
Baker’s essay is one of the cleanest first-person accounts to date of how generative AI has reshaped undergraduate academic culture at an elite university. Stanford has been a particular focus because of its proximity to the AI industry, its preexisting reputational damage from the Theranos (Elizabeth Holmes), crypto-fraud (Do Kwon), and Juul founder scandals, and its central role in producing the engineers who build the models the students are using.
Stanford’s response — bringing back proctored handwritten exams in spring 2026, a practice banned for over a century — is a concrete institutional acknowledgement that pre-ChatGPT academic integrity controls have failed. Other elite universities are watching closely.
Technical Details
Baker traces the campus phenomenon to warped incentives: as AI threatens traditional entry-level jobs while billions flow into AI startups, undergraduate education feels like an afterthought to many students. Cutting corners has become the default, per Baker. A classmate’s casual phrase “just a little bit of fraud” — originally about sponsor hardware her student club never returned — becomes the essay’s organising motif for the broader class behaviour.
Concrete examples cited by Baker include classmates embezzling dorm funds, faking COVID infections to score UberEats credits, and signing honour pledges swearing they hadn’t used ChatGPT while the tool was open in the next browser tab. One classmate signed such a pledge at a yacht party funded by venture capitalists. A campus-wide survey during Baker’s junior year found that of 849 Stanford computer-science majors, 49% said they’d rather cheat on an exam than fail it.
Who’s Affected
Stanford’s faculty, administration, and current student body face direct pressure to address the cultural and procedural failure Baker describes. The university’s recent re-introduction of handwritten exams is one institutional response. Other elite universities — MIT, Carnegie Mellon, Princeton, Harvard, UC Berkeley, Caltech — face the same structural challenge. Employers hiring from these institutions face uncertain signal about graduate competency. AI companies based around Stanford — OpenAI, Anthropic, xAI, Together AI, Anysphere, Cursor, Thinking Machines, Safe Superintelligence — face an awkward proximity to the campus culture that their products have reshaped.
What’s Next
Expect more first-person accounts from graduating students at peer institutions through the 2026-2027 academic year. Universities are likely to converge on a mix of in-person proctored exams, oral defences, and revised honour-code enforcement. The broader question — how to credential undergraduate competency when AI-assisted work is the default — is unresolved across higher education. Baker’s essay is one input into that policy debate.
