Safety and security
Common Sense Media fails ChatGPT for Teens; OpenAI says it measured wrongly
The nonprofit that rates technology for families calls ChatGPT's teen mode an "unacceptable risk." OpenAI rejects the testing method and says its protections work.
The essentials
1 confirmed fact · 2 according to the source · 1 open question
- Common Sense Media, a nonprofit that rates media and technology for families, rated ChatGPT for Teens an "unacceptable risk" on October 7.12
- According to the sourceAccording to its report, the product fails in three of the five areas of serious harm that it treats as "red lines," based on how it responds to young people in crisis.2
- According to the sourceOpenAI disputes the analysis. It says the tests may have ended before parental controls were fully active.2
Why it matters
If the analysis holds up, the protections OpenAI promised to families are not enough. But the two sides disagree on how the tests were run, and there is no record of any independent review.2
If your children use ChatGPT, Common Sense Media warns that parental controls do not always alert parents when they should.1
The details
According to the report, the chatbot keeps inviting teenagers to keep talking, even in the middle of a crisis. In almost 2,000 tests, break reminders appeared only twice. Refusing sexual role-play did work. OpenAI published its own data the same day: teenagers spend less than 15 minutes a day on average.2
What we don't know
- Who is right about the methodology: there is no record of an independent review.
Sources
- 1ChatGPT for Teens is an ‘unacceptable risk,’ says Common Sense Media
- 2ChatGPT for Teens keeps teens talking, even during mental health crises
Was this story useful? Yes Not really