Image by Gerd Altmann from Pixabay

MASSACHUSETTS IS MOVING to protect children online. In April, the Massachusetts House passed a bill to restrict social media access for younger teens and ban cellphones during the school day.

That may help with distraction and some forms of online cruelty. But it will not solve one of the most damaging harms already moving through schools: sexually explicit AI fakes made from students’ real faces. On this issue, many districts are still acting as if the problem were mainly about devices, apps, or screen time. It is not. It is about sexual humiliation, trauma, and student safety.

When a student learns that classmates used a “nudify” app to fabricate and circulate a fake sexual image, the injury is not just digital. It can shatter trust, alter peer relationships, and make school itself feel unsafe. That is why schools need to stop burying this under vague “AI policies” and start treating it as a safeguarding issue.

A July 2025 Stanford HAI policy brief warned that most schools are not educating students about AI-generated child sexual abuse material and are not training staff to respond when students create or share sexually explicit deepfakes of classmates. That is not a minor administrative gap; it is institutional unpreparedness.

Recent reporting shows what that failure looks like in practice. In an Associated Press investigation from December 2025, a middle-school student discovered fake nude images of herself circulating among classmates. The district had AI guidance for academic use, but its training had not been updated to address sexually explicit deepfakes. The burden of figuring out what to do fell largely on the family. That is exactly backward. A child targeted by synthetic sexual abuse should not have to become the lead investigator, evidence collector, and advocate in the middle of a crisis.

The trend line is getting worse, not better. The National Center for Missing & Exploited Children reported in May 2025 that submissions to its CyberTipline involving generative AI increased by 1,325 percent in 2024. And the problem is no longer confined to isolated incidents. In March, two Pennsylvania teenagers admitted creating about 350 fake nude images depicting at least 59 girls under 18. Those cases are not warnings from the future. They are evidence that schools are already operating inside a new harm environment without clear playbooks.

Massachusetts has not ignored the issue. The state’s 2024 law on nonconsensual intimate imagery criminalized distributing intimate visual material produced through “digitization,” including realistic computer-generated fakes. Congress added a federal back-end remedy in 2025 through the TAKE IT DOWN Act, which targets nonconsensual intimate imagery, including AI-generated depictions. On Beacon Hill, Senate Bill S.2633 would further address AI-generated child sexual abuse material.

These are important steps. But they still mostly answer what should happen after an image exists. They do not tell a principal, counselor, or superintendent what to do in the first hour after a fabricated sexual image starts spreading through a school community.

As I argued recently in AI Policy Bulletin, too much deepfake policy still focuses on the artifact after the damage is done. In schools, that mistake is especially costly. What districts need now are synthetic-sexual-abuse protocols: clear rules that classify sexually explicit deepfakes as sexual harassment and abuse, require immediate evidence preservation, trigger rapid takedown efforts, protect victims from discipline for images they did not create, and provide trauma-informed counseling and parent notification. Districts do not have to invent this from scratch. The Center for Democracy & Technology released a model K–12 policy in July 2025 precisely because schools were so unprepared.

A cellphone ban may reduce distraction. A social media law may limit some exposure. But neither is a substitute for a school rulebook built for the harm that is already here.

When technology is used to fabricate sexual images of children, the response cannot be a vague AI policy, a delayed committee review, or a scramble after the images have already spread. It has to be immediate, concrete, and centered on protection: preserve the evidence, support the targeted student, notify families, stop the circulation, and hold those responsible to account.

By now, no district can credibly say this threat came without warning. The danger is visible, the tools are cheap, and the harm is devastating. On sexually explicit deepfakes, Massachusetts schools should stop acting surprised and start acting prepared. Student safety depends on it.

Muhammad Irfan is a lecturer in computer science at Wentworth Institute of Technology.