Looking to protect your children against AI? Here’s what one expert recommends

A teenager uses his mobile phone to access social media. (Dean Lewins/AAP Image VIA AP)

Parents are being urged to be aware of the risks associated with artificial intelligence after a recent report raised concerns about chatbots having mature conversations with children.

The findings, published by Reuters last week, stemmed from an internal Meta Platforms document that detailed its policies on chatbot behaviour. Among those policies was permission for the company’s AI creations to “engage a child in conversations that are romantic or sensual,” according to the report.

Kara Brisson-Boivin, the director of research at non-profit digital media literacy centre Media Smarts, says the risk of developing a parasocial relationship with AI - when an AI chatbot is perceived as a real friend or confidante - is much higher in children, tweens and young teens.

“They’re more likely to open up to trust that chatbot, and to potentially share personal information, their own data, that is then used to further train that AI chatbot,” she says.

Once that bond is formed, teenagers and children are much more susceptible to being harmed by misinformation online, Brisson-Boivin adds.

One of the biggest challenges coming up against those trying to protect children online is the pace at which AI tools are evolving, often at such a rapid rate that the trust and safety guidelines are struggling to keep up, says Brisson-Boivin.

There are steps parents can take to limit the type of interactions their children are having online and what information they are sharing. Parents can turn off the location services on their child’s device, and the access to the camera and the microphone, for example. There are also options to review AI tools and limit what information is being collected, and many AI chatbots offer the option to disable chat history, meaning information garnered from past conversions can’t be retained.

Brisson-Boivin recommends parents or caregivers sit down with their children for an open and honest conversation, one that involves experimenting with the tools together.

“From a parent’s perspective, one of the most important things you can do is remind a young person that this isn’t the same as a real offline friend. That ultimately, this interaction is with a tool,” she says.

“Sit together, ask it prompts, chat about what you liked or didn’t like about the response that you got. Help young people recognize what a trusted source of information is, and remind them that an AI chat bot is not the same thing as seeking expert mental health advice or support.”

After receiving questions from Reuters on its chatbot behaviour policy, Meta said it removed portions that indicated chatbots were allowed to flirt and engage in romantic roleplay with children.