ChatGPT Age Prediction Launched: New Restrictions for Minors
ChatGPT Age Prediction: New Teen Mode & Restrictions Explained
 |
| Image: Emiliano Vittoriosi / Unsplash |
Today marks a pivot in AI safety as OpenAI launches this new behavioral analysis system. Its primary goal? To shield users under the age of 18 from graphic violence, sexual roleplay, and other harmful material. While the famous chatbot previously initiated age-restricted content based on simple age gates, this new model is far more effective, and invasive. It analyzes account behavior, such as typing patterns, usage times, and topic choices, to identify users who might be minors. This means that even if a user tries to bypass the age gate by entering a fake birthdate, they are still catchable. The system now prioritizes "digital fingerprints" over the user's self-reported age.
OpenAI explicitly detailed the types of digital fingerprints the model is trained to detect. The company notes that it looks at "how long an account has existed, typical times of day when someone is active, usage patterns over time, and a user’s stated age." The AI model uses these inputs to continuously train itself for better accuracy.
Here is how the behavioral analysis breaks down:
- Time of Use: The model flags heavy usage during typical school hours (roughly 8 AM – 3 PM).
- Language Patterns: It detects specific slang, simplified sentence structures, or trending teen vernacular.
- Topic Selection: Frequent queries related to homework, teen-centric viral challenges, or gaming can trigger the system.
If the AI determines these signals align with a user under 18, it silently switches the account to "Teen Mode," regardless of the birthdate originally entered during signup. It is important to consider that this is a backend change; you may not receive a notification until you attempt to generate content that is now restricted.
If your account is flagged by the Age Prediction model, you will lose the ability to generate specific types of content. The new "Teen Safety" filters are aggressive and aim to block:
- Sexual Roleplay: Any attempts at romantic or erotic roleplay (NSFW) are immediately blocked.
- Graphic Violence: Detailed descriptions of gore, injury, or combat are filtered out.
- Self-Harm Depictions: Any content that discusses or visualizes self-injury or eating disorders is prohibited.
- Beauty Standards: Content promoting unhealthy dieting or extreme physical body standards is restricted.
These restrictions are hard-coded into the "Teen Mode" and cannot be bypassed with "jailbreak" prompts.
However, we know that AI models can hallucinate. Since this model relies on behavioral analysis, there is a chance it might wrongly flag an adult account as a minor. This is most likely to happen to users who casually prompt or, most importantly, to gamers. If you find your access restricted, you cannot simply change your birthdate in the settings. You must go through a formal identity verification process.
But don’t worry! OpenAI provides a way to remove the flag from your account by verifying your identity. OpenAI has partnered with Persona, a secure identity verification platform, to handle these appeals.
How to Verify Your Age:
- Go to Settings > Account in ChatGPT.
- Look for the "Age Verification" status. If you are flagged, it will say "Restricted" or "Teen Safety On."
- Click "Verify Age."
- You will be redirected to Persona to scan a driver’s license or passport.
- You must take a real-time photo to prove you match the ID.
Alongside these new restrictions, parents are still able to implement further guardrails through parental controls. This new model was introduced following serious concerns and incidents regarding the platform's impact on minors.
