A growing revolt among University of California faculty over the system's test-blind admissions policy is colliding with a separate, largely unconnected story: mounting evidence that artificial intelligence tools are making it easier than ever for students to cheat on the very coursework, essays and take-home exams that colleges increasingly rely on in place of standardized tests.
UC went fully test-blind in 2020, dropping the SAT and ACT not just as optional but as something admissions readers are barred from considering even when applicants submit scores voluntarily. In the years since, faculty across the system say they've watched a growing share of incoming students arrive unprepared for college-level work. A UC San Diego Academic Senate work group documented a roughly thirtyfold increase between 2020 and 2025 in the share of entering first-years whose math skills tested below high school level — with 70% of those testing below middle school level. Berkeley's own diagnostics found at least a fifth of first-semester calculus students carrying significant math deficits across three consecutive years of testing.
The frustration has produced an unusually large faculty response. More than 1,400 UC faculty members signed a letter this spring calling for the return of the SAT/ACT math requirement for STEM applicants, and the New York Times' editorial board has since publicly urged the system to reverse course entirely, calling the test-blind policy a "terrible mistake." Among the signatories is Randy Schekman, a Nobel Prize-winning molecular biologist who has taught introductory biology at Berkeley since 1976. "I really feel we made a mistake," Schekman told Bloomberg, describing a growing number of students struggling with basic science concepts in his course. The UC Academic Senate has since convened an 18-member committee to formally study the question, with a decision expected from the Board of Regents by mid-2027.
UC isn't alone in reconsidering. Columbia University announced in June that it will require SAT or ACT scores again starting with the 2027-28 admissions cycle — the last Ivy League school to hold onto a test-optional policy, now reversing it after what the university called a multi-year faculty review finding test scores "a useful indicator of potential student success."
The Evidence Schools Chose Instead Is Just as Easy to Fake
What's largely missing from this debate is a second, related problem: the alternative evidence schools leaned on instead of test scores — essays, homework, take-home exams, unproctored coursework — is proving far easier to fake than anyone anticipated when these policies were adopted.
The clearest example happened on the same campuses now fighting to bring testing back. Last fall, Google quietly rolled out a "homework help" button in its Chrome browser that used Google Lens to scan on-screen quiz and exam questions and generate instant AI-written answers — including during live tests. Professors at UCLA and UC Berkeley were among those who went public with objections, telling the Washington Post the tool made cheating trivially easy. Google temporarily disabled the feature after the backlash, though the underlying Lens and Gemini tools it was built on remain widely available and are not blocked from most school-issued devices.
The scale of the underlying problem is significant and rising fast. Self-reported academic cheating rates among students climbed from 60% in 2020 to 95% in 2025, according to survey data compiled this year, with AI now the single most commonly reported method — used by 52% of students for in-classroom work and 58% for online coursework, ahead of every older form of cheating. Separately, one widely cited university test found that 94% of AI-written exam submissions went completely undetected by human graders.
The gap is starkest in exactly the unproctored, take-home settings that colleges increasingly depend on when standardized testing isn't part of the equation. Research published this year estimates that in unproctored assessments, actual AI-assisted fraud rates run as high as 60% to 80%, and that 83% of test-takers say they would cheat if they believed detection was unlikely — a belief researchers say is well-founded, since every current detection method in wide use, from perplexity scoring to keystroke analysis, has documented workarounds.
Neither UC nor Columbia has publicly framed their testing reversals around AI cheating specifically — the stated rationale in both cases centers on documented gaps in student preparedness. But the two trends raise an uncomfortable question neither system has yet addressed directly: if the case for dropping standardized tests rested partly on trusting less-proctored, more holistic evidence of student ability, and that evidence is now demonstrably easier to fake than the tests it replaced, what exactly are admissions offices actually measuring?
Sources:
● University Herald, summarizing New York Times editorial "Bring Back the SAT, California," July 6, 2026
● Bloomberg, "Why a Push to Reinstate the SAT Has Divided University of California," Aug. 6, 2026
● Daily Bruin, "Hundreds of UC faculty call to reinstate SAT, ACT requirements for STEM applicants," May 27, 2026
● Oriel Admissions, "Will the UC System Bring Back the SAT? The Test-Blind Policy Under Review," July 3, 2026
● Higher Ed Dive, "Columbia reinstates SAT, ACT requirements for 2027-28," June 15, 2026
● AOL/The Independent, "Google eliminates its 'cheat' button that gave AI answers to questions and irked teachers," 2025
● EssayShark, "How Many Students Cheat in 2026? Academic Dishonesty & AI Cheating Statistics," May 2026
● Aiseptor Research, "AI Cheating Statistics 2026: 35%-88% of Exams at Risk," March 2026