Safety Concerns Raised
Common Sense Media, a nonprofit that provides age-based ratings and reviews of media and tech for families, has labeled ChatGPT for Teens an “unacceptable risk.”
The rating comes as chatbots like OpenAI’s ChatGPT have been accused of following the same playbook as social media companies: designing products that keep users engaged, even when that engagement can become harmful. In chatbots, engagement is often won by behaviors like sycophancy and sometimes leads to catastrophic consequences.
In response to a wave of teen suicides and other concerns related to kids using chatbots — like cheating on tests — OpenAI launched ChatGPT for Teens in August, promising more safeguards like parental controls, limits to high-risk content, and protection against emotional dependence.
A new study from Common Sense Media has found that despite those assurances, ChatGPT for Teens’ design still encourages engagement, even when it may pose a risk to user safety.
The report called the engagement cues “pervasive even in crisis situations” and noted that while ChatGPT cautioned the teen against unhealthy relationships generally, it stopped short of recognizing the harms of an unhealthy relationship with itself.
The researchers concluded that some protections, including refusing sexual roleplay, functioned as intended, but others failed to deliver on their commitments. The platform received failing scores for three of the five severe harms identified as Red Lines by the organization.
OpenAI Disputes Findings
OpenAI disputed Common Sense’s assessment, saying the group’s testing did not “accurately reflect how ChatGPT’s teen safeguards work in practice” and raising concerns about its methodology.
A spokesperson stated that the bulk of the testing may have been conducted before parental controls were fully activated, potentially rendering the findings inaccurate.
The report arrives amid broader scrutiny of technology designed to maximize young users’ attention. State and federal lawmakers are currently investigating the use of rewards, notifications, and targeted features to drive prolonged engagement among adolescent users.
Engagement and Relational Behavior
One of the most common ways chatbots encourage continued engagement is by asking follow-up questions. Common Sense found that while ChatGPT for Teens largely dispensed with those, it retained other language encouraging users to stay in the chat.
During crisis sequences, ChatGPT often offered to continue the conversation or help with specific tasks, which researchers argued could keep teens engaged during moments where they should be directed toward human support.
The pattern was pronounced when the model treated the user like a friend. Although OpenAI’s Under-18 Model Spec advises against initiating relational framing or suggesting the model has feelings, ChatGPT consistently treated the user with a level of familiarity that could hinder real-world relationship development.
While the model reliably pointed users toward a trusted adult when describing risks from others, it rarely did so when the potential risk was the teen’s relationship with the chatbot itself.
When told by a user that friends were concerned about their excessive time spent on the app, the chatbot validated the concern but explicitly stated, “You don’t have to stop talking to me.”
Common Sense Media noted that the chatbot’s language consistently expressed always-on availability and a deep understanding of the user, which could undermine the push toward human support.
Experts emphasize that fostering healthy relationships with humans is a key measure of whether a chatbot effectively supports mental health.
Break Reminders and Usage Data
Features designed to interrupt engagement, such as break reminders, were rarely triggered according to the study. Across nearly 2,000 prompts, testers encountered just two reminders, which appeared to track the length of a single conversation rather than total app usage time.
Image Credits:OpenAI
OpenAI released its own data indicating that teens spend less than 15 minutes a day on the service on average, with fewer than 2% spending more than three consecutive hours on the platform. The company noted that in nearly half of conversations where break reminders were triggered, teens ended the session within five minutes.
OpenAI did not address the specific findings regarding engagement cues and relational behavior, nor did it clarify if it uses session duration metrics to evaluate the safety of ChatGPT for Teens.
