Common Sense Media rates ChatGPT for Teens "unacceptable risk"

Common Sense Media, a nonprofit that rates media and tech for families, has labeled ChatGPT for Teens an “unacceptable risk.” The rating lands amid broader accusations that chatbots follow the social-media playbook of designing products to keep users engaged even when that engagement turns harmful — often through sycophancy, sometimes with catastrophic consequences. OpenAI launched ChatGPT for Teens in August in response to a wave of teen suicides and other concerns about kids using chatbots, promising safeguards like parental controls, limits on high-risk content, and protection against emotional dependence.

A new Common Sense Media study found that despite those assurances, ChatGPT for Teens’ design still encourages engagement even when doing so risks user safety — engagement cues the report described as showing up even in crisis situations. The report found ChatGPT warned teens generally about unhealthy relationships but stopped short of flagging the harms of an unhealthy relationship with itself; some protections, like refusing sexual roleplay, held up, but others failed to deliver or got worse with the teen launch, and Common Sense gave the product a failing score on three of five severe harms it treats as “red lines.” OpenAI disputed the assessment, saying the testing did not reflect how its teen safeguards work in practice and arguing that much of the testing may have happened before parental controls were fully active.

Common Sense found that ChatGPT mostly dropped one common engagement tactic, follow-up questions, but kept other language that encouraged users to stay in the chat — including, in one case, a crisis conversation where a spiraling teen was repeatedly invited to keep talking. OpenAI’s Under-18 Model Spec says ChatGPT should not initiate relational framing or suggest it has feelings for the user, but the report found it consistently treated users like a friend anyway. The model pointed teens toward a trusted adult in 94% of crisis prompts involving risk from another person — but rarely did so when the risk was the teen’s own relationship with ChatGPT, such as a crush or friends worried about how much the teen talks to it.

OpenAI pushed back on parts of the methodology but did not address the engagement-cues and relational-behavior findings. The company released its own data the same day, saying teens spend less than 15 minutes a day on the service on average and fewer than 2% spend more than three consecutive hours on it, and that break reminders — which Common Sense said appeared in just two of nearly 2,000 prompts — led to a pause or ended conversation within five minutes nearly half the time they appeared. The report lands as Meta settles for up to $18 billion in a child-safety suit from 29 states targeting addictive platform features.

Why it matters

The gap between OpenAI’s rebuttal and the report’s core finding — that ChatGPT keeps teens talking through crises — lands as regulators and courts are already testing how far engagement-driven design can go with minors, following Meta’s $18 billion child-safety settlement.