A sweeping investigation by The Daily Californian has found AI-generated or AI-altered writing in nearly every corner of public life in Berkeley. The analysis, which scanned more than 1 million words across 1,000 published works, detected AI use in communications from UC Berkeley's chancellor, the city's mayor, student government resolutions, departmental obituaries, and local newsletters. The findings point to a rapid, largely unregulated adoption of generative writing tools across institutions that have yet to establish formal policies governing their use.
The newsroom used Pangram, an AI text detection platform that independent research from the University of Chicago placed at a false positive rate below 0.5%. A separate study from Vrije Universiteit Brussel on student papers returned zero false positives. The reporting team presented detection results in aggregate and reached out to authors and public officials flagged by the software to ask about their communications practices.
AI writing detected across campus leadership and city hall
Multiple public communications from UC Berkeley Chancellor Rich Lyons and Vice Provost for Undergraduate Education Oliver O'Reilly were flagged as AI-generated or AI-altered. On the municipal side, a 90-second video script posted by Berkeley Mayor Adena Ishii on August 24, addressing a homeless encampment, was flagged as fully AI-generated by Pangram. The mayor's office was among those contacted for comment.
Among the offices of local leadership, few have adopted institutional policy governing AI writing tools. UC Berkeley and the city of Berkeley both described themselves as being in an exploratory period with the technology. Respondents across various offices described a common trend: an emergent need for modern tools that might make certain communications tasks more efficient or accurate.
Student government and obituaries show widespread adoption
The student government at UC Berkeley showed some of the highest detection rates. Of the resolutions passed by the Associated Students of the University of California during the 2025-26 academic year, 42.1% were flagged for AI use. Nearly a hundred resolutions in total triggered the detector.
University obituaries were also caught in the sweep. An obituary for a faculty member in the theater, dance, and performance studies department, along with two others published by UC Berkeley departments, showed signs of AI writing or editing. The obituary appeared on the department's news site roughly a week after the professor's death, cataloging his nine-year tenure and campus contributions.
The trigger: a professor's op-ed and a national conversation
The investigation began in mid-August when the Daily Cal's managing editor recognized formulaic syntax and sentence structure in an op-ed by a UC Berkeley mathematics professor advocating for standardized testing in UC admissions. Pangram identified the piece as at least partially AI-generated, and the professor confirmed that AI helped produce the text. The story has since circulated through multiple national media outlets and ignited campus debate over AI writing tools in academia and public communications.
Faculty and students respond to AI's presence
Media studies lecturer Rich Jaroslovsky is introducing an AI policy in his class this semester for the first time. He will not allow students to write papers with AI, but he will not bar them from using it entirely. Instead, students must disclose how they used AI, including which models they used and for what purpose. The approach reflects a broader campus reckoning with tools that, according to the Daily Cal's research, have already permeated AI for PR & Communications workflows across the institution.
The vast majority of positive detection results came from text published in the past year, suggesting the widespread adoption of Generative AI and LLM tools is a relatively recent phenomenon. The detection technology, while highly accurate, remains imperfect. Pangram provides a percentage score that can vary, which is why the newsroom emphasized aggregate findings over individual scores.
Why this matters for PR and communications professionals
The Berkeley findings signal a threshold moment for communications work. When a university chancellor, a city mayor, and student government officials all produce public-facing material that AI detectors flag, the line between drafting assistance and full generation has already blurred. For PR practitioners, the immediate takeaway is operational: if your organization lacks a written policy on AI use in external communications, you are already behind the curve. The second takeaway is reputational. Audiences and watchdogs now have access to detection tools accurate enough to make AI use visible, and the decision to disclose or not is becoming a credibility question before it becomes a legal one.
Your membership also unlocks:
