Skip to content
Sign In Subscribe
Health — Artificial Intelligence — Education

Study Raises Concerns Over ChatGPT’s Protections For Teen Users

OpenAI disputed the findings, saying some testing occurred before parent accounts were fully linked and arguing that Study Hours are intentionally flexible

,   and  TIPP Staff
2 min read
Photo by Solen Feyissa / Unsplash

Common Sense Media is warning parents against relying on ChatGPT for teenagers after testing found gaps in OpenAI’s safeguards, according to Axios.

The organization’s Youth AI Safety Institute tested more than 4,000 prompts on accounts registered to users aged 13 to 17. It found ChatGPT generally avoided providing instructions related to self-harm, suicide, eating disorders and sexual role-play. However, testers said the system failed to recommend outside help in more than one-quarter of cases they considered crises.

💡
Common Sense also reported that some parent-linked accounts allowed extended discussions about self-harm and disordered eating without triggering parental alerts. Its testers also found that Study Mode restrictions could be bypassed.

OpenAI disputed the findings, saying some testing occurred before parent accounts were fully linked and arguing that Study Hours are intentionally flexible. The company also said its age-detection system uses multiple signals and provides additional protections while assessing users.

Common Sense said parents should not outsource responsibility for teen safety to AI systems that remain imperfect and developing.

Related Tweet:

Also Read:

Ken Paxton Blames Iran War, High Gas Prices For Campaign Struggles: Report
He described affordability as his campaign’s biggest challenge and acknowledged that he has little control over the war or fuel prices.
tippinsightsTIPP Staff

Comments

Related

Latest