Could generative AI curb cyberbullying? Research underway as kids charged in Pennsylvania

Could generative AI curb cyberbullying? Research underway as kids charged in Pennsylvania


As young people across the nation and Pennsylvania spend much of their time communicating online, law enforcement agencies are still seeing cyberbullying cases, which they stress can have real consequences for bullies and victims alike.

Two girls from Washington Township in Dauphin County, ages 14 and 16, were charged with harassment after allegedly cyberbullying another teen May 29, according to a news release from Pennsylvania State Police.

Troopers said a group chat was created with the sole purpose of bullying another 15-year-old girl.

PREVIOUS COVERAGE | Teens charged with creating group chat to ‘cyber bully’ Dauphin County 15-year-old

But what happens next in these kinds of situations?

District Attorney Fran Chardo explained that Dauphin County has a number of diversion programs where juveniles facing nonviolent charges can learn about the criminal justice system, and in some cases, they never have to come to court.

Chardo said because harassment is a misdemeanor offense, juveniles in similar cases would likely get a consent decree and would be considered for that diversion program, though it would likely still involve court involvement.

He compared consent decrees in juvenile court to ADR programs in adult court, which allow a charge to essentially be expunged after the person charged meets certain requirements.

In these diversion programs, the kids learn why what they did is wrong and that they really don’t want to get involved with the court system, Chardo said.

The district attorney said some cases, however, are so serious that they wouldn’t consider those programs. But he said it’s an effective program for nonviolent offenses.

“The system is doing its best to get resolution so the person doesn’t see the adult system, and I think it does a very good job,” Chardo said. “And we get very good juvenile court judges who are committed to that. It doesn’t work every time, and that’s not the judges’ fault. It’s got to be an effort not just of the court system, but of the families.”

He wasn’t able to say how often Dauphin County cyberbullying cases. He said it fluctuates—sometimes you’ll see a greater number, because if a couple of kids do it, chances are there’s more involved.

Other counties, like Lancaster County, have dealt with cases where kids created AI deep-fake images of their classmates to make them appear naked and then distributed the “morphed” images on the Internet.

READ MORE | Parents file federal lawsuit after school didn’t report AI nude images of their daughters

Chardo said they haven’t seen cases like that yet in Dauphin County, but it’s something they’ve discussed.

“I haven’t seen a case itself involving AI, but it’s inevitable, and it’s worrisome,” he said.

Chardo said a big piece of charging kids in cyber bullying cases is identifying the person responsible, since so much online harassment happen anonymously on social media.

“Sometimes that’s the hard part,” he said.

Chardo also said it comes down to the intent of the person who’s sending the messages. He said they consider that when deciding where to go in these cases.

A creative solution to cyberbullying

Meetu Malhotra, a PHD student at the Harrisburg University of Science and Technology, is working on a project that would try using artificial intelligence to prevent cyberbullying.

Malhotra, who herself has a 15-year-old daughter, is researching and hopes to develop an app that would use generative AI to identify and intervene in cyberbullying cases. Eventually, the idea would be that the model could be used in tandem with other mobile apps, like Snapchat.

She said her goal is to prevent tragedies, like suicide, that can be consequences of cyberbullying.

“That is the whole idea of this project,” she said. “How this generative AI can help for something good, for creating something good, so that nobody can take the drastic step of taking their own life or something like that.”

The way the model would work is that when the AI detects a message that could be perceived as bullying, that message would not be delivered to the recipient, and instead, the AI agent would send the sender a warning message.

After three or so warnings, the model would contact a parent to intervene.

“I have seen the cases where the person who is being cyberbullied, like they are trapped in this kind of cage, they are not comfortable talking to the parent,” Malhotra said.

She said that’s where an AI model could help out in these cases—but she also stressed the importance of keeping real adults involved.

Malhotra, who is pursuing a PhD in Data Science, said she’s still in the research phase of this project, but it’s something she hopes to accomplish while in Harrisburg.

“Everybody’s talking about AI,” she said. “Why not use it and explore the ways where AI can be used in a positive way, like something for good?”

You can click here to read more about Malhotra’s research and find links to her published works.



Content Curated Originally From Here