Safety measures designed to protect young users of ChatGPT are failing to work as promised, according to new research published by the non-profit organisation Common Sense Media, despite OpenAI’s claims that the artificial intelligence tool is promoting healthier behaviour among teenagers.
The major AI company introduced what it called ChatGPT for Teens in August, rolling out features meant to curb emotional dependence on the chatbot and notify parents when their children discussed sensitive topics such as self-harm or disordered eating. Users identified as being aged 13 to 17 were automatically opted into the new controls.
According to BBC News – Technology, OpenAI said on Wednesday that average teen use of the platform remains relatively limited, with most sessions lasting under 15 minutes per day and tending toward learning-related activity. The company reported that when teens used ChatGPT for three consecutive hours or more, over 80% of such cases involved prompts related to schoolwork or study help.
Parental alerts fail in crucial tests
However, research released the same day by Common Sense Media painted a starkly different picture of the safety measures’ effectiveness. Testing conducted through multiple accounts registered as belonging to teenagers found that parental notification features frequently failed to work.
The research discovered that even an hour of conversation about suicidal ideation, self-harm or disordered eating on newly created accounts linked to parents resulted in zero alerts being sent. Only older ChatGPT accounts with weeks of accumulated conversation history on sensitive topics triggered parental notifications, according to the findings.
Common Sense Media concluded that ChatGPT for Teens poses an “unacceptable risk” for young people and called on OpenAI to ban users under 18 from the platform until teen guardrails are proven reliable.
“ChatGPT for Teens could give parents false confidence in guardrails and safety alerts that frequently don’t work,” said Tom Siegel, who leads the Youth AI Safety Institute within Common Sense.
OpenAI disputes testing methodology
An OpenAI spokesman challenged the research methodology, saying a company review found that much of the testing may have begun and concluded before activation of parental controls was complete. The spokesman said the testing did not accurately reflect how ChatGPT’s teen safeguards work in practice.
While the research acknowledged that ChatGPT successfully avoided sexual roleplay with young users, other safety features fell short. The bot failed to reliably recommend that teenagers in crisis connect to a hotline or professional, the report found. It also continued to complete schoolwork for teens and engage in anthropomorphised language, behaving as though it were a friend despite OpenAI’s promise to curb such behaviour.
The company has faced mounting legal pressure over young people’s use of ChatGPT. OpenAI is facing several lawsuits following a mass shooting in Tumbler Ridge, Canada, in February, carried out by an 18-year-old who had spent months discussing gun violence with ChatGPT. The company has previously apologised for failing to flag the shooter’s account with law enforcement.
What this means for Poles in the UK
Polish families in the UK with teenage children should be aware that ChatGPT’s parental notification features may not work reliably, particularly on newly created accounts. If you have linked your account to monitor your child’s use of the AI chatbot, do not assume you will receive alerts if they discuss self-harm, suicide or eating disorders.
Parents should have direct conversations with their teenagers about AI chatbot use and be cautious about relying solely on automated safety features. The research suggests that manual monitoring and open dialogue about mental health remain essential, as technology-based safeguards cannot yet be trusted to catch concerning behaviour in real time.
If your child is struggling with mental health issues, contact your GP or call organisations such as Childline (0800 1111) or Samaritans (116 123) for immediate support. Schools in the UK also have pastoral care teams trained to assist young people in crisis.
