During a separate session featuring Little Rock School District, an audience member recounted his conversation with a vendor who offered varying levels of data protection at different price points.
“They’re gambling with kids’ PII,” he said. “It can’t happen like that.”
Arkansas, for its part, requires vendors to sign agreements specifying how they'll handle student data, which helps make the vetting process of tools simpler.
But this isn’t the case for every district across the country, and among even the smallest districts, attitudes toward data security and AI were clear: Data privacy and compliance must be verified and enforced with every new AI tool.
Policy for AI Use Must Be a Moving Target
In 2022, Santa Margarita Catholic High School patched together a policy about how teachers and students should use AI. The policy required citation but gave no guidance on what that meant. It didn't explain AI's limitations, biases or how the technology worked, and it was largely restrictive.
In 2023, the high school began experimenting more deeply with teacher use cases. It launched a monthly newsletter about AI developments and began providing basic literacy resources to students, teachers and parents. Despite progress, the school soon encountered a new obstacle.
"We drank the AI tool Kool-Aid," said Jorge Ledezma, director of educational technology. “We thought AI was going to take care of grading and other things." When these tools underdelivered, trust eroded. Simultaneously, the school began using AI checkers to “monitor rather than mentor” students using AI. These checkers often contradicted one another, leading to contentious conversations.
WATCH: Industry experts discuss AI’s 2026 trajectory.
2024 and 2025 marked a turning point. Santa Margarita created an AI task force that met monthly and included faculty, staff, students and parents. The body developed a system that doesn’t strictly prohibit AI use among students; the work must be their own, but they may use AI for assistance as long as they disclose it.
For teachers, it became clear that AI use cases varied by department. For example, Gordon Minton, a math teacher at Santa Margarita, noted that AI underperforms at creating math problems. Instead, Minton said, he focuses on using AI to digitize content stored as PDFs.
Ledezma said he also realized that AI can explain its own capabilities and limitations and provide important teaching moments for students. For example, he used Google Gemini to explain tokens. AI does not have infinite memory, which means it starts to lose context in longer conversations, leading to diminishing returns over the course of the interaction.
“Schools should use this moment to get clear on fundamental questions.” Ledezma said. “What, really, is education? What, really, is plagiarism? What does it mean to have authentic human relationships, versus interactions with bots?"
Click the banner below to learn how artificial intelligence training can modernize classrooms.