States and school districts across the country are tightening rules on artificial intelligence as the 2026-27 K-12 school year begins, with education leaders citing concerns over student learning, screen time and the risks of rapidly advancing technology.

As concerns about AI’s impact on K-12 education have grown, dozens of states have introduced or passed legislation aimed at regulating its use in schools.

California

In California, Gov. Gavin Newsom has taken steps to restrict AI use as it has become a cause for concern. This past week, Newsom signed Senate Bill 813 and Assembly Bill 1405, which aim at keeping AI accountable through safeguards and auditors.

“The concerns raised in recent incidents reinforce what California has long recognized: artificial intelligence holds extraordinary promise, but it must be developed and deployed with meaningful safeguards to protect the public,” Newsom said in a statement, even calling on the federal government to implement national safeguards.

The Center Square reached out to the Democratic governor's office for further comment on what this looks like in the K-12 classroom but did not receive a response.

The Los Angeles Unified School District told The Center Square that the Generative AI Ad Hoc Committee met in early September to develop guardrails and procedures for students' use of AI. The district said it allows students 13 and older to access approved generative AI tools if they complete Digital Citizenship Instruction and acknowledged the Responsible Use Policy. The district said since the beginning of the 2026-27 school year, it has restricted students' access to generative AI platforms while it reviews safeguards.

In an interview with The Center Square, Lance Izumi, senior director of the Center for Education at the Pasadena, Calif.-based Pacific Research Institute, said the lack of long-term research on AI's impact on student learning should warrant caution.

“Since AI is such a new resource, there is still no definitive research on the long-term impact of AI on student learning. In particular, there is no long-term tracking of students and their learning and performance using AI tools,” Izumi told The Center Square.

“Given the lack of research data, slowing the introduction of AI into the classroom makes sense and gives policymakers the time to better understand how AI can help and hurt student learning, and then craft policies accordingly,” he added.

Izumi said AI could provide benefits when used for tutoring or research, but Izumi raised concerns about cheating and the potential decline in students' development of fundamental skills.

“On the benefit side, AI can be helpful as a tutoring tool that meets students where they are academically and provides them with explanations, practice, and assessments to improve their learning,” Izumi said. “However, there is a dark downside to AI as well. Teachers are complaining that AI allows students to cheat on virtually any assignment that they give them.”

"Research also shows that AI users performed worse than non-AI users on standardized test essays. AI usage can also become a crutch, with students failing to develop key skills such as reading and writing,” Izumi said.

New York City

New York City officials have also moved to limit AI use in schools.

New York City Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels launched a moratorium to combat artificial intelligence use in the classroom, along with a screen time policy for the 2026-27 school year.

The city's approach also includes an AI literacy module for high school students twice a year. According to the Mayor’s Office, the modules will cover “AI-related bias, risks, ethical considerations and impacts on careers and future skills.”

Arizona

In Arizona, lawmakers are considering legislation that would require schools to teach students about artificial intelligence.

The Arizona House Committee on Artificial Intelligence and Innovation passed House Bill 4005, which would require school districts and charter schools to provide instruction on the “ethical, moral and educational uses” of AI, The Center Square previously reported.

Under the legislation, school districts and charter schools would be required to submit an annual report to the Arizona Department of Education by Oct. 15 describing the type of AI instruction they provide to students.

The Center Square reached out to Arizona Superintendent Tom Horne for comment, but he was unavailable.

Illinois

The Illinois State Board of Education launched an AI guidance package in June 2026, providing school districts with guidance on evaluating and using artificial intelligence products.

The roughly 400-page AI Blueprint emphasizes that AI should support teachers and student-teacher relationships.

“The guidance is not organized around artificial intelligence as a technology to be adopted for its own sake. It is organized around the enduring educational premise that the teacher-student interaction is the centerpiece of teaching and learning,” the document states.

“AI may assist with planning, access, feedback, communication, operations, and resource development, but it must remain a means to inform teaching and learning rather than a substitute for educator judgment, student relationships, or the developmental and civic purposes of public schooling,” the document said.

The Center Square reached out to ISBE for further information but did not receive a response.

“It is important to tell students early and up front about both the positive and negative impacts that AI can have on their education,” Izumi told The Center Square. “We must be honest with our children, especially when they are young, and not put the biases of adults ahead of the interests of students. And we have to get students to buy into how they can best maximize their own learning. Limiting AI usage, when supported by research evidence, does not mean that students can't learn about the technology itself.”

“AI policy in education should be guided by research data showing how AI benefits or harms the learning of students, so ideally research components should be attached to AI policies so that these policies can be validated or invalidated,” he added.