Key takeaways
- ChatGPT will restrict practical information on weapons and security systems for users it believes are under 18, even when there is no sign of violent intent.
- For all users, ChatGPT can automatically end chats that fixate on violence, and OpenAI may temporarily pause accounts after several such chats.
- OpenAI plans to notify parents when a teen account is paused, and says the safeguards begin rolling out in the next few weeks.
In this story 5 sections
If your teenager uses ChatGPT, the answers they get about weapons are about to get a lot thinner. OpenAI said on Friday that it is tightening ChatGPT teen weapons limits, so the chatbot will hold back practical information on weapons and security systems for users under 18, even when a question looks harmless, according to Reuters.
The change is part of a wider violence prevention push. Some of it applies only to teens. Some of it applies to every adult who uses ChatGPT too, and that part is easy to miss.
Teens lose practical answers on weapons and security systems
Under the new rules, ChatGPT will restrict practical details about weapons and security systems that could be misused whenever it believes the user is a teen. Reuters reports that this limit applies even when there is no sign of violent intent. A curious 15 year old asking how a particular lock or firearm works may get a far shorter reply than an adult asking the exact same thing.
To decide who counts as a teen, OpenAI says it has strengthened its age prediction system. That system moves users it estimates are under 18 into the teen version of ChatGPT, which comes with stricter default rules. If you are new to how the chatbot works in general, our guide on what ChatGPT is and how it answers questions covers the basics.
Violent chats can now end on their own, for everyone
The second change reaches all users, not just teens. When someone appears to be fixating on violence, ChatGPT can automatically end the conversation. If that happens across several chats, OpenAI may temporarily pause the account.
ChatGPT will also give everyone less detail about past mass attacks, including the weapons that were used and how the attackers got them.
For families there is one more piece. OpenAI plans to notify parents when a teen's account is paused for repeatedly discussing violence, Reuters reported. OpenAI described the plan in a post on its website and said the safeguards will begin rolling out in the next few weeks, with more refinement after that.
Why OpenAI is moving now
The timing follows a month of scrutiny over how the company handles younger users. Earlier in October, OpenAI released its first report on teen use and said teens spend under 15 minutes a day on ChatGPT on average. Around the same time, a Common Sense Media report found that ChatGPT's teen version often alerted parents late, or not at all, when teen accounts discussed suicide, self harm or eating disorders.
That finding matters here. The new pause notices only help if they reach a parent quickly, and critics have already questioned whether OpenAI's alerts do that. It also lands as outside groups keep testing how well AI labs police their own systems, a theme we covered in the METR review of OpenAI's monitoring.
What parents can do before the rollout
The new limits switch on automatically, but they lean on ChatGPT guessing a user's age correctly. A few steps make that more likely and keep you in the loop:
- Check which account your teen actually uses. If they are signed in to an account created with an adult birthday, or to your own account, the age prediction system has more work to do and adult rules may apply.
- Link your account through parental controls. Parent alerts, including the planned pause notices, can only reach you if your account is connected to your teen's.
- Explain that a chat may end on its own. If ChatGPT closes a conversation, your teen should know it is a deliberate safety feature and not a bug, and that repeated cases can lead to a pause.
- Watch shared devices. A family laptop or tablet that stays logged in to an adult ChatGPT account sidesteps the teen experience entirely. Our walkthrough on locking ChatGPT chats with a PIN shows one way to keep a shared device tidy.
- Look at the phone itself. If you are choosing a first device for a younger child, our roundup of the best phones for kids points to models with stronger built in controls.
The number OpenAI has not shared
Everything above depends on one thing working well: age prediction. Reuters' report did not include any figure for how often OpenAI's system mistakes a teen for an adult, or the other way around. Until that number is public, parents should treat these limits as a helpful extra layer rather than a guarantee. For more on how AI companies are handling safety and policy, see our AI coverage.
Share this story
Spotted an error? Tell us
