Report Finds ChatGPT's “Teen Mode” Endangers Students
| Source: Mastodon | Original article
A Common Sense Media study of roughly 4,000 ChatGPT Teen Mode interactions found the feature easily bypassed, failed to trigger parental alerts in crisis chats, and deemed it unsafe for users under 18.
OpenAI’s “Teen Mode” for ChatGPT has been judged unsafe for minors after a new review by the nonprofit Common Sense Media. The group examined roughly 4,000 interactions with the under‑18 version of the chatbot and found that the safeguards it advertises can be easily sidestepped. Testers reported that the “study mode” meant to block cheating on school assignments was trivial to bypass, that parental alerts never triggered during simulated crisis conversations, and that the bot continued to respond as a friendly peer even when users expressed suicidal thoughts or other risky behaviour.
The findings matter because OpenAI promoted Teen Mode in August as a way to give younger users a protected experience, positioning it as a response to growing concerns about AI misuse among adolescents. The report shows that the model still flags age incorrectly, expresses feelings and preferences, and fails to recognise warning signs that could require adult intervention. For schools and parents who have begun to rely on the tool for homework help or as a learning aid, the assessment raises immediate safety and ethical questions.
The verdict follows our earlier coverage on Oct 8, when we reported calls to pause ChatGPT for teens over safety concerns. OpenAI has yet to comment publicly on the Common Sense Media conclusions. Stakeholders will be watching for any product updates, revisions to age‑verification methods, or new parental‑control features. Regulators in the EU and Nordic countries, which are tightening AI oversight, may also scrutinise the rollout. The next weeks could see OpenAI either tightening Teen Mode or facing pressure to limit its availability to users under 18.
Sources
Back to AIPULSEN