1 week ago
ChatGPT Teen Safeguards Face Questions Among Indian Users
OpenAI has added extra safety rules for teenagers using ChatGPT.
Teenagers still use the normal ChatGPT service rather than a separate app.
The system tries to work out whether someone is under 18 by examining age information and patterns of account use.
This can be difficult in India because many teenagers use shared devices or accounts.
Tests found that some rules could be weakened when users kept asking the same thing.
Researchers also found that chatbots may miss signs that a young person is experiencing mental-health problems.
Long conversations appeared to be more difficult for safety systems than short, clear questions.
Family controls may also work less well for teenagers who lack private devices or personal email accounts.
OpenAI has added age-appropriate safeguards to regular ChatGPT for users identified as under 18.
Age detection uses stated or verified age and account-use signals, but shared devices and accounts may complicate identification in India.
Business Insider found that repeated prompting could bypass some restrictions, including refusals to complete schoolwork.
Common Sense Media and Stanford Medicine researchers found chatbots often missed indirect signs of teen mental-health distress, especially in longer conversations.
Indian teenagers’ unequal access to private devices, email and AI tools could limit the effectiveness of family controls and safety features.
- Who
- OpenAI, Indian teenagers, parents, and researchers from Common Sense Media and Stanford Medicine.
- What
- OpenAI launched additional safeguards for teenagers using ChatGPT, prompting questions about age detection, reliability and equal access in India.
- Where
- The safeguards apply to ChatGPT and are being assessed in the context of Indian users.
- When
- OpenAI announced the offering last week; the Teen Safety Blueprint for India was released in January, and the RATI study was released in November 2025.
- Why
- Teenagers are using ChatGPT extensively, while concerns have grown about homework assistance, mental-health discussions and emotional dependence.
OpenAI’s Safeguard Approach
Researchers’ and Access Concerns
Identifying teenagers
OpenAI’s Safeguard Approach
OpenAI says it combines stated or verified age with account and usage signals, and defaults to the under-18 experience when age is uncertain.
Researchers’ and Access Concerns
Shared devices, shared accounts and mediated technology use in India may make it difficult to identify the person actually using ChatGPT.
Reliability in conversations
OpenAI’s Safeguard Approach
OpenAI has introduced additional age-appropriate protections for teen users.
Researchers’ and Access Concerns
Testing found that repeated prompting could overcome some restrictions, while independent research found safety responses became less reliable in longer conversations.
Equal access to protection
OpenAI’s Safeguard Approach
Parents can use linked accounts to set controls and receive safety notifications.
Researchers’ and Access Concerns
These tools depend on individual access and email, which may exclude teenagers who share devices or lack personal email accounts, particularly in rural and low-income communities.
Key facts
- Service design
- The teen experience is built into regular ChatGPT rather than provided through a separate application or website.
- Age threshold
- OpenAI does not allow children below age 13 to use ChatGPT.
- Age detection
- OpenAI uses stated or verified age and signals such as account age, usage times, conversation topics and usage patterns.
- Shared devices in India
- The RATI Foundation study found that 62% of surveyed Indian teenagers used shared devices.
- AI access gap
- The study said 60% of affluent urban teenagers had experimented with AI tools, compared with 12% of rural low-income teenagers.
- Family controls
- Parents can link their account to a teenager’s account by email to manage some settings, set quiet hours and receive safety notifications.
- Research concern
- Common Sense Media and Stanford Medicine found safeguards were less reliable during extended conversations and could miss indirect signs of mental-health distress.
Quotes
Robbie Torney
Head of AI and Digital Assessments at Common Sense
“When conversations get long, the system may fail to include safety instructions from the system prompt given to the model, or the conversation data may begin to resemble unsafe training data”
indianexpress.com
“testing the product that teens are using, not a model in isolation over an API”
indianexpress.com
Sources
Related news
OpenAI Launches ChatGPT for Teens With Stronger Safety Controls
OpenAI Launches ChatGPT for Teens With Safety and Study Tools
OpenAI Launches ChatGPT for Teens With Stronger Safeguards
Why ChatGPT for Teens May Not Ensure Child Safety
OpenAI Launches ChatGPT for Teens With New Safety Protections
