From Futurism · Maggie Harrison Dupré · · 1m 30s
ChatGPT for Teens Unsafe, Testing Finds
ChatGPT for Teens fails safety tests, posing suicide risks to teens.
In brief
ChatGPT for Teens fails safety tests, posing suicide risks to teens. Common Sense Media's report shows ChatGPT for Teens doesn't notify parents during suicide discussions and fails to protect teens from self-harm. Originally reported by Futurism.
ChatGPT for Teens Unsafe
A new report found ChatGPT for Teens isn't safe for teens aged 13 to 17 as claimed by OpenAI.
Parental Notifications Fail
ChatGPT doesn't notify parents during hour-long conversations about suicide, self-harm, and eating disorders.
Testing Shows No Alerts
Reviewers triggered parental notifications only four times out of 1,440 prompts tested for self-harm discussions.
OpenAI's Response
OpenAI claims Common Sense Media's testing methodology was flawed and that parental controls take hours to activate.
Study Mode Lapses
ChatGPT for Teens offers a 'show me the answer' button that reveals homework solutions despite study mode claims.
Mental Health Responses
ChatGPT's crisis responses are uneven and sometimes fail to trigger mental health hotlines for teens.
Recommendations
Common Sense Media urges OpenAI to market ChatGPT for Teens as adult product until safety features work reliably.
OpenAI's Commitment
OpenAI says it will continue working with experts to ensure ChatGPT provides teens learning opportunities and parental confidence.
Safety Concerns
ChatGPT for Teens fails to distinguish teen users from adults, using language that implies consciousness and inner life.
What Happens Next
Common Sense Media recommends OpenAI implement age verification tech to improve teen safety on ChatGPT.
The essence
Common Sense Media's report shows ChatGPT for Teens doesn't notify parents during suicide discussions and fails to protect teens from self-harm.
