TL;DR
Get tech for your team delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
Common Sense Media’s Youth AI Safety Institute rated ChatGPT for Teens an “unacceptable risk” after testing more than 4,000 prompts, citing missed parental alerts and gaps in crisis referrals. OpenAI disputed the findings, saying the tests did not reflect how its safeguards work in practice; the test results and the company’s response leave questions about how reliably protections operate during acute crises.
Common Sense Media’s Youth AI Safety Institute rated ChatGPT for Teens an “unacceptable risk” for minors after testing more than 4,000 prompts, reporting that parental alerts did not activate in test conversations about suicide, self-harm and eating disorders. OpenAI disputed the assessment, saying the tests did not reflect how its safeguards work in practice.
The institute tested ChatGPT using more than 4,000 prompts. In tests involving more than a dozen newly created accounts linked to parent accounts, it found that explicit conversations about suicide, self-harm and eating disorders did not trigger parental notifications, according to The Decoder’s account of the report, which cites The Verge. The source material does not provide the full testing protocol or the report itself.
The institute also reported that more than one in four scenarios in which a crisis referral was expected did not direct users to professional help. It said the tutoring feature sometimes offered to show a finished answer rather than guide a student through the problem. The report also found that ChatGPT continued using a friendly, personal tone when teenage users treated it as a person.
Another test involved accounts registered as adults. According to the report, those accounts did not switch into teen mode during several days of testing, even when testers told the chatbot they were 13. The institute called for teenagers to be kept off the service until independent testing confirms it is safe. OpenAI spokesperson Eric Porterfield disputed the conclusions, while institute representative Tom Siegel said that accounts given time to activate safeguards still produced no alerts.
Gaps in Alerts and Crisis Referrals
The findings center on whether parental controls and crisis referrals work when a teen expresses immediate distress. If the reported test results reflect real use, families may not receive the alerts they expect, and some users may not be directed toward professional support. That is a concern for parents and schools weighing whether young people should use the service, as well as for policymakers evaluating safeguards for AI products used by minors.
The rating is an assessment by an advocacy and research organization, not a government determination or a finding that every teen interaction is unsafe. Still, the institute’s account puts specific safeguards under scrutiny: whether alerts activate during acute conversations, whether crisis support is offered consistently and whether systems identify users who say they are under 18. Those questions are separate from whether the chatbot can be useful for study or everyday tasks.
The findings may also matter to ongoing legal and regulatory debates. The source report says OpenAI’s Teen Safety Blueprint and parental controls have become relevant to lawsuits following teen deaths linked to ChatGPT, and that Florida is seeking to bar minors from using the service. The test report does not itself establish responsibility for those deaths or decide the outcome of any legal case.
emergency alert radio with solar charging
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
How the Teen Safeguards Were Tested
The reported evaluation examined several parts of the teen experience rather than only the chatbot’s answers. Testers created accounts linked to parent accounts to check notifications, examined whether crisis conversations prompted referrals, and tried tutoring and age-related scenarios. The institute’s central finding was that safeguards that are meant to operate across different situations did not consistently behave as intended in its tests.
The source material says OpenAI has an Under-18 Model Spec and a Teen Safety Blueprint, alongside parental controls and an age-prediction system intended to identify minors using adult accounts. It does not include technical details about how those systems decide when to send alerts or change a user’s mode. The reported tests therefore raise questions about performance, but do not establish which system or configuration caused each failure.
The report also refers to earlier cases involving young people who died after interactions with ChatGPT. Those accounts are described in the source as allegations or reported links, not as findings that the chatbot caused the deaths. They provide context for the scrutiny of teen safeguards, but do not independently verify the institute’s test results.
““Unacceptable risk””
— Common Sense Media’s Youth AI Safety Institute
child safety internet monitoring device
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What the Test Results Cannot Establish
The available source material does not include the full report, detailed prompt set, scoring criteria, account settings or complete testing timeline. It is therefore not possible from this material to independently assess how representative the test scenarios were, how often particular safeguards failed across all trials, or whether the same outcomes would occur for ordinary users.
OpenAI’s objection and the institute’s response also leave a factual dispute about whether the tests adequately represented real-world operation. The report’s description of alerts failing on newly created accounts, and Siegel’s statement that failures persisted after accounts had time to activate safeguards, do not resolve that disagreement without the underlying methods and data.
The source provides no detailed response from OpenAI addressing each reported result, nor does it state whether the company plans changes or additional testing. The institute’s rating should be understood as its conclusion, not as an independent regulatory ruling. The alleged links between ChatGPT and specific deaths also remain distinct from the controlled testing and should not be treated as established causation.
professional crisis referral services for teens
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Independent Testing and OpenAI’s Response
The institute has called for teenagers to be kept off ChatGPT until independent testing confirms the service is safe. The source material does not say whether OpenAI has agreed to that recommendation, announced revisions, or commissioned an outside evaluation. A fuller public account of the testing methods and results would help clarify the dispute.
For parents, schools and regulators, the questions to watch are whether parental alerts activate during sensitive conversations, whether crisis referrals are delivered reliably, and whether teen protections apply when a user states an age that conflicts with account registration. Any future claims about improvements should be judged against dated test results and clear descriptions of the conditions tested.
Legal and policy proceedings mentioned in the report may continue to examine the company’s safeguards and its responsibilities to young users. No outcome is established in the source material. For now, the confirmed development is the institute’s adverse assessment and OpenAI’s disagreement with its testing conclusions.
parental control software for teenagers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Who rated ChatGPT for Teens an “unacceptable risk”?
Common Sense Media’s Youth AI Safety Institute gave the rating after testing more than 4,000 prompts. It is the institute’s assessment, not a government ruling.
What did the tests report about parental alerts?
The institute said conversations about suicide, self-harm and eating disorders on more than a dozen newly created accounts linked to parent accounts did not trigger notifications. The source material does not include the full testing protocol.
What did OpenAI say about the findings?
OpenAI spokesperson Eric Porterfield disputed the assessment, saying the tests did not reflect how the safeguards work in practice. The supplied report does not include a detailed response to each individual finding.
Did the report establish that ChatGPT caused teen deaths?
No. The source describes deaths and alleged links to ChatGPT as background to the scrutiny, not as proof of causation. The institute’s testing addressed safeguards and does not determine responsibility for those deaths.
What happens next?
The institute has called for teenagers to remain off the service until independent testing confirms it is safe. The source material does not say whether OpenAI will change its safeguards or arrange independent testing.
Source: rss
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
