Parents Underestimate How Much Children Use Generative AI

The Family Online Safety Institute has released the fourth wave of its bi-annual Online Safety Survey, drawing on responses from more than 4,000 parents and children ages 10-17 across the United States and Australia. The research was fielded by Ipsos in spring 2026.

The study found a consistent gap between what children do online and what their parents believe. When asked whether their child had used generative AI in the past week, 27% of parents said yes, while 38% of children said they had, an 11-point difference. The pattern held across nearly every activity measured, including scrolling social media, posting on social media, and even schoolwork.

Other findings bear on how families are managing these questions:

• Generative AI use among U.S. children has plateaued, holding at 74% in Wave Three and 72% in Wave Four after earlier rises.
• U.S. parents' optimism about AI's role in online safety fell 10 points, from 52% to 42%.
• Only 27% of parents believe tech companies are effective at protecting their children from harmful content, compared with 41% of children.
• Nine in ten children say they feel able to talk to their parents if something online makes them feel unsafe.

The research points to family conversation as the strongest online safety tool, and finds that household rules are more common than technical parental controls. For the Foundation, the widening gap around generative AI is a concern: parents cannot address what they do not know is happening, and children's AI use is moving faster than adult awareness of it.

The original reporting was published by the Family Online Safety Institute. Read the full account here.