Radnor school district is looking to enlist outside experts to prepare for future deepfake incidents
Radnor drew criticism after a so-called deepfake incident last year that targeted freshman girls. Schools have been forced to grapple with inappropriate deepfakes as students use generative AI.
The Radnor Township School District is preparing for a possible repeat of a so-called deepfake incident last year that targeted freshman girls, looking to enlist outside help on how to better support victims.
At a school board policy committee meeting Tuesday, administrators presented updated regulations around bullying and harassment, including that using artificial intelligence to create or share “sexualized, indecent, demeaning or intimate content involving another person” was prohibited.
The new rules — which follow policy changes passed last month — also say the district will consider who should be present while interviewing students, in light of a student’s age, sex, and background and the issues being investigated. Parents had complained about many aspects of the district’s handling of last year’s incident, including that girls victimized by deepfakes were interviewed alone by male administrators.
But more changes are possible, administrators said Tuesday. Scott Hand, the district’s director of technology innovation and instructional design, said Radnor “will be engaging outside professionals beginning this summer into fall” to review its policies and practices.
Some board members also suggested the district retain outside help to be ready for another incident.
“The statistics say, something like this will happen again,” said Susan Stern, who asked administrators to consider hiring “a resource we could have on hand.”
“We’re going to work really hard to educate our kids that it doesn’t happen here in Radnor, but our best response is to be prepared,” Stern said.
Schools have increasingly been forced to grapple with deepfakes as students use generative AI to manipulate photos of their peers — sometimes with tools that can convert clothed photos into naked ones. Pennsylvania has criminalized explicit AI-generated images as child sexual abuse material, and another measure advancing in Harrisburg would require schools to immediately notify law enforcement about AI incidents.
Radnor drew intense criticism from some victims’ parents for its handling of the deepfakes, which came to light in December and resulted in a student being charged with harassment.
District officials said no images were ever discovered. But parents said the district had failed to adequately investigate boys’ reports of pornographic videos and downplayed the incident in community messages, further harming victims.
Parents also accused the district of failing to initiate a Title IX harassment investigation, arguing it was wrong to characterize the incident as “off-campus.”
Radnor parents are organizing a panel discussion Thursday in King of Prussia, with deepfake victims from Radnor and Lancaster Country Day School, advocates, and state officials, including Attorney General Dave Sunday, describing risks and how schools can better respond.
The rules presented to Radnor’s policy committee Tuesday specify that the district has authority over off-campus behavior when it “materially and substantially disrupts the operations of the school” — a common legal standard.
Jannie Lau, the committee’s chair, said that given how online kids are, any online harassment would seem to meet that threshold.
“That’s my takeaway — it would be extremely rare for this conduct to happen, and for us to come to the conclusion it doesn’t impact our operations," Lau said.
Lau said the district should consider hiring an outside law firm to conduct future Title IX investigations, to ensure more impartiality.
“It’s just human nature — you don’t want to bite the hand that feeds you,” Lau said, noting that a district solicitor might have an interest in siding with the district, to avoid jeopardizing that relationship.
In comments Tuesday, Audrey Greenberg, a parent of one of the victims, said she appreciated the rule changes, but that they didn’t go far enough. If a student reports AI-generated sexual content, she said, the district’s Title IX coordinator and central administrators should be notified immediately.
She also said the description of “off-campus” behavior was too vague — saying deepfakes harm students’ attendance, friendships, and mental health — and argued that the rules wouldn’t prevent a poor administrative response.
“It was so painful to get those emails calling the event rumors and speculation, when a crime occurred,” Greenberg said, referring to messages from Superintendent Ken Batchelor, including one that described the deepfakes as images made “to move and dance.”
“For crying out loud, this is pornographic material,” Greenberg said.
Stern, the board member, said earlier in the meeting that Radnor would be dealing with the fallout for a long time.
“I don’t think that there’s anyone in this district who will ever hear the words ‘generative AI’ again, and not feel a little tug,” she said. “We had children who were harmed.”
