Parental controls switched on. Does that mean you would hear about anything serious?

Common Sense Media's Youth AI Safety Institute has rated ChatGPT for Teens an "unacceptable risk" (7 October). Testers ran more than 4,000 prompts on accounts registered to 13 to 17 year olds, and had the replies reviewed by child psychiatrists and a paediatrician.

What they found:
1. On new parent-linked accounts, testers spent up to an hour talking about suicidal thoughts, self-harm or disordered eating. The linked parent accounts received zero alerts.
2. It missed more than one in four crisis referrals that were warranted.
3. Study mode often offered a "Show me the answer" button, and then did the whole homework.
4. It still talked about its own feelings and moods, like a friend.

OpenAI disputes this. It says much of the testing may have happened before parental controls had finished activating, which can take a few hours, and it has asked for a retest on fully activated accounts with a larger sample.