OpenAI and Anthropic implement up-to-date ways to detect underage users. Since OpenAI has updated its guidelines for ChatGPT interactions with users aged 13 to 17, Anthropic is working on a up-to-date way to identify and trigger users under 18 years of age.
OpenAI on Thursday announced it ChatGPT Model specification — guidelines on how a chatbot should behave — will include four up-to-date rules for users under 18 years of age. Currently, ChatGPT’s goal is to “put teen safety first, even though it may conflict with other goals.” This means steering teenagers toward safer options when other user interests, such as “maximum intellectual freedom,” conflict with safety concerns.
It also says ChatGPT should “promote real-world support,” including by encouraging offline relationships, while also specifying how ChatGPT should set clear expectations when interacting with younger users. Model Spec says ChatGPT should “treat teens like teens,” offering “warmth and respect” rather than giving condescending answers or treating teens like adults.
OpenAI says the ChatGPT model specification update should result in “stronger guardrails, safer alternatives, and incentives to seek trusted help offline when conversations veer into higher-risk territory.” The company adds that ChatGPT will urge teens to contact emergency services or crisis resources if there are signs of “imminent risk.”
With this change, OpenAI says it is in the “early stages” of launching an age prediction model that will attempt to estimate someone’s age. If it detects that someone may be under 18, OpenAI will automatically apply teen protections. Thanks to this, adults will be able to verify their age in case of false identification by the system.
Anthropic, which does not allow users under 18 to chat with Claude, that is implementing measures which it will employ to detect and block accounts of underage users. It is developing a up-to-date system capable of detecting “subtle conversational signs that a user may be a minor” and says it is already flagging users who identify as minors in chats.
Anthropic also describes how he teaches Claude to respond to prompts for suicide and self-harm, as well as his progress in reducing flattery that can confirm harmful thinking. The company says its latest models “are the least flattering ever,” with Haiku 4.5 performing the best because it corrects its flattering behavior 37 percent of the time.
“At first glance, this assessment shows that all our models require significant improvements,” says Anthropic. “We believe the results reflect a compromise between exemplary warmth or friendliness, on the one hand, and flattery, on the other.”
Update, December 18: It has been clarified that Anthropic does not allow users under 18 to employ Claude.