The Journal

For the better part of 70 years, the machines that profile us and the machines we confide in ran on separate tracks; with conversational AI, the watcher became the confidant.

A return, five years on, to a question I could not properly answer at my PhD viva: if privacy is fundamentally about the formation of the self in relation to others, what does it mean when the population most affected — adolescents — is the population in active formation?

The trial proved that social media companies built a product that is addictive by design; that is the supply side.

Jonathan Haidt's argument is somewhat incomplete: anxiety among young people had been climbing steadily since at least the 1950s, and pre-existing anxiety predicted heavier and more compulsive phone use, not the other way around.

Relief, I have spent the last several years learning, is not the same thing as safety.

Children's most disruptive behaviours — the meltdowns, the refusals, the aggression — are usually not failures of self-control.

Most adult conversations about keeping children safe end up being about restriction.

What happens to a young person's sense of self when much of their daily experience is shaped by systems that learn from them in real time?