California’s public colleges and universities are spending millions on high-tech solutions to catch plagiarism and artificial intelligence (AI) misuse, but mounting evidence reveals these investments may be rife with technological blind spots, privacy risks, and questionable educational value. As higher education worldwide faces a new era of “AI arms race,” the experience of the California system demonstrates how quick fixes may fall short, fueling new debates that resonate in classrooms from Bangkok to Berkeley.
The prominence of AI writing tools like ChatGPT has sparked widespread concern among professors who question whether students are still the authors of their own work. Within months of this technological leap, software companies rushed to sell tools promising to detect both plagiarism and AI-generated writing. California State University alone spent an additional $163,000 in 2025 to license Turnitin’s AI-detection upgrades, pushing total annual costs above $1.1 million. In all, Cal State campuses have directed a stunning $6 million to Turnitin since 2019, and dozens of community colleges and University of California campuses have signed similarly expensive contracts, according to a collaborative investigation by CalMatters and The Markup (calmatters.org).
Why does this concern Thai students and educators? Thailand, like California, is rapidly digitizing its education sector in the wake of the Covid-19 pandemic. Online learning and digital assignments have become the norm, and with them comes the temptation—to both students and institutions—to rely on automated surveillance and plagiarism detection tools. Understanding the problems California faces now offers a stark preview of potential pitfalls for Thai colleges, which stand on the cusp of similar investments.
At the center of the controversy is Turnitin, a well-known company that has evolved from simple plagiarism checks to advanced, yet deeply flawed, AI-content detection. Turnitin’s business model requires colleges to grant it “perpetual, irrevocable, non-exclusive, royalty-free, transferable and sublicensable” rights over millions of student papers. According to contract terms, these student works are then pooled into Turnitin’s vast database—now approaching 1.9 billion papers—which the company uses to develop and market its services.
But as the technology advances, so too do its limitations. Turnitin’s detectors have been criticized for false-flagging genuine student work as AI-generated, while sometimes missing true instances of AI-supported cheating. Their algorithms, built to recognize stylistic markers common in machine-written text, often confuse the work of non-native English speakers or students using simple grammar tools like Grammarly for full-fledged AI authorship. In one case investigated, a student at San Bernardino Valley College, who was a native Spanish speaker, received a zero after Turnitin flagged her work as AI-generated, despite her insistence that the writing was entirely her own.
These experiences are hardly isolated. A Center for Democracy & Technology survey found one in five high schoolers knew someone wrongly accused of AI cheating. Another study by Common Sense Media showed Black teens were doubly likely to be unfairly flagged—a troubling echo of longstanding patterns of unequal discipline in American schools.
Expert opinions increasingly challenge the efficacy and ethics of such software. A senior director of AI programs at Common Sense Media argued that the tens of millions spent on licenses would be better directed toward professor training and clear institutional guidelines. He observed, “It’s probably better to invest in training for professors and teachers, and also creating frameworks for universities to message to students how they can and can’t use AI, rather than trying to use a surveillance methodology to detect AI in student writing.”
Meanwhile, many instructors report that the most telling signs of AI misuse—such as fabricated sources and “hallucinated” quotes—escape current detectors. As an English professor at College of the Canyons noted, the tool is “not good at catching” such hallmarks, instead flagging stylistic elements or even proper quotation as suspicious. With repeated false positives, honest students and those struggling with linguistic barriers are increasingly “caught in the middle,” dealing with the resulting stress and uncertainty.
This fractured atmosphere stands in stark contradiction to what many educators now see as the most effective deterrent against academic dishonesty: trusting relationships between students and teachers. Data shows that, despite new technology, rates of cheating have remained largely unchanged, challenging the premise that technological surveillance serves as a meaningful deterrent (Jesse Stommel, University of Denver). Instead, critics argue, these tools may erode student morale and stoke a culture of suspicion, with unintended consequences for mental health and educational engagement.
The industry built around these fears is lucrative. Since 2014, College of the Canyons’ costs for Turnitin soared from about $6,500 to nearly $47,000 a year. Across California, public records show upwards of $15 million spent since 2018 on Turnitin licensing alone. The Los Rios Community College District has paid almost $750,000; Los Angeles Community College District’s single-year license reached $265,000. UC Berkeley, meanwhile, is locked in a 10-year contract totaling nearly $1.2 million. When contract terms changed after Turnitin acquired a competitor in 2018, colleges were forced to accept global, perpetual data mining of their students’ intellectual property or go without—an option few chose, despite rising privacy concerns.
The educational arms race shows no sign of slowing. As digital assignments and AI bot usage become ever more embedded in student routines, some academics now openly recommend their institutions forgo third-party detectors. Stanford University, for example, shifted away from Turnitin, citing the technology’s tendency to “erode feelings of trust and belonging among students and raise questions about their privacy and intellectual-property rights.”
For Thai educators, administrators, and policymakers, these lessons are crucial. Thailand’s universities already use plagiarism detection for research and increasingly incorporate digital learning assignments. Should they follow California’s example, the consequences—expensive, sometimes faulty, and privacy-invasive solutions—should be carefully weighed against the possibility of fostering trust, training, and clear communication within the academic community.
Historical and cultural lessons must also play a role. Thailand, with its strong emphasis on educational harmony and teacher-student respect, may be uniquely positioned to open broader conversations about appropriate technology integration. Allowing students to contribute meaningfully to policies governing AI and digital assignments could counteract feelings of surveillance and encourage responsible, innovative uses of technology in learning. Similar approaches have been explored in Scandinavian and Japanese educational models, which favor collaborative trust over adversarial monitoring (OECD Education Policy Outlook).
Looking ahead, the rise of generative AI in education is inevitable. Chatbots are already used across Thailand for customer service, language study, and research assistance. But the debate raging in California highlights that technological quick fixes can spiral into new problems. Both research and practical experience point toward the need for continuous faculty development, student guidance, and equitable frameworks that acknowledge the blurred lines between digital assistance and dishonesty.
Ultimately, the most practical advice for Thai education stakeholders is to invest in a well-rounded digital literacy curriculum. This encompasses critical thinking, academic writing standards, and an understanding of how AI and detection tools work—all supported by robust teacher-student relationships. Parents and students should be aware of how their digital assignments might be monitored or stored, and universities should fully inform all parties when adopting new agreements with foreign technology vendors.
As universities worldwide grapple with the same dilemmas, Thailand stands at a crossroads: whether to invest in imperfect, costly surveillance or to build the adaptive, trust-based culture that best fosters genuine academic progress in the AI age.
For detailed analysis see the investigative report by CalMatters and The Markup (calmatters.org), and recent findings from Common Sense Media (commonsense.org), the Center for Democracy & Technology (cdt.org), and the OECD’s Education Policy Outlook (oecd.org).