Artificial intelligenceHow the new ChatGPT parental control works
OpenAI announced the new parental feature for more safety for children and young people at the end of September 2025. This announcement was certainly also a reaction by the company to the legal proceedings initiated in the USA and the associated negative reporting. The accusation was that ChatGPT endangered children and young people with inappropriate content and exacerbated mental health problems such as suicidal thoughts by always agreeing with users. A few months after the announcement, OpenAI has now also implemented the security settings in Germany so that they can be used by users in this country.
According to ChatGPT's General Terms and Conditions, the minimum age is 13 years. Children under the age of 13 are therefore not allowed to use the program or create an account. Young people between the ages of 13 and 18 need their parents' consent to use the program. As with most online services, ChatGPT does not yet have a reliable age check or parental consent check. It can therefore be assumed that children under the age of 13 also use ChatGPT and that many parents have not actively consented to their children between the ages of 13 and 18 using the service.
We have explained why platforms specify a minimum age in their terms and conditions in this article.
How to activate parental controls on ChatGPT
Both the child and the parent must create an account with ChatGPT in order to use the parent support service.
- Click on your profile picture.
- Click on "Settings".
- Click on "Parental control".
- Click on "Add family member".
- Enter the email address used to register the account you wish to link.
- Select the appropriate field. Either "My parent or legal guardian" or "My child".
- Then click on "Send".
- As soon as the other person has agreed, a pop-up message will appear. Click on "Apply" again to finally connect the accounts.
What functions does ChatGPT's parental control offer?
Once the parent and child accounts have been successfully connected, parents can make settings. The most important setting options are
Quiet times
Here you can set a time period during which ChatGPT cannot be used (e.g. 22:00-06:00). Disadvantage: Only a single time period can be set. It is therefore not possible, for example, to set different quiet times for school hours and at night. It is also not possible to differentiate between different days of the week (e.g. school days and weekends).
Reduce sensitive content
With this setting, OpenAI promises that young people will be shown "(...) less graphic content, viral challenges, sexual, romantic or violent role-playing games and extreme beauty ideals (...)". We were unable to test how reliably this security setting works.
Group chats
Here parents can prevent their children from being invited to group chats within ChatGPT or from creating them themselves. If this function is not deactivated, anyone who knows the child's account name or linked email address can invite the child to a group chat with up to 20 people. It is also possible to join via an invitation link.
Image generation
Here parents can decide whether or not their children are allowed to generate images with ChatGPT.
Improve the model for everyone
If this function is deactivated, the data entered by the child will not be used to train and improve the AI model, according to OpenAI. ChatGPT also works fully without this function. There is therefore no reason why parents should not allow their child's data to be processed.
In addition to the setting options described above, there are several other parental control options for other OpenAI services. For example, they can make settings for the AI video generator Sora and the web browser ChatGPT Atlas.
Is ChatGPT's parental control useful and effective?
First of all, it is pleasing that, after several years, there are setting options for parents at all. It would be better if services such as OpenAI provided effective safety settings for the protection of children and young people from the outset and did not only react after massive criticism and public pressure.
Some of the functions that have now been added are certainly useful, but they are often not very flexible and therefore not equally suitable for all families. A good example of this is the definition of rest periods: Here it would be desirable if more than one time slot could be set.
Comparable parental control functions often offer a function that shows both parents and children how long the child is allowed to use the device. This provides a basis on which parents and children can jointly decide whether usage is within normal limits and whether certain rest periods would be appropriate. However, such a function is not available in the ChatGPT parental controls.
It is also questionable whether the content filter is actually effective. It is known from other AI platforms that they continue to output sensitive and illegal content despite supposedly functioning filters(e.g. Character.ai). In this context, OpenAI only talks about "reducing" sensitive content and not about preventing it completely. Parents should therefore not rely too much on this setting to protect their children from all inappropriate content in ChatGPT.
Parents who use ChatGPT's parental controls have no insight into their child's communication with the chatbot. However, OpenAI promises to inform parents if the conversation with the chatbot could indicate a dangerous situation. In this case, the chat messages will first be investigated by OpenAI employees. According to OpenAI, they are also working on informing the police, for example, in the event of a dangerous situation. How exactly this works, whether this function is already active and whether it will also be used in Germany is unfortunately not known. Nevertheless, it certainly makes sense to let children know from the outset that their conversations with the chatbot are being monitored and can trigger a report to parents or even the police.
In our test, the child account was able to delete the connection to the parent account without the parent account being informed. According to the information provided by ChatGPT, this should not be possible at all. Although children can unilaterally disconnect from the parent account, the parent account should at least be informed. Unfortunately, we cannot say whether our case is an exception or whether the parental account notification actually does not work. However, parents should bear in mind that their children could switch off parental controls without their consent and use ChatGPT without restriction.
How to support your child in dealing with AI
Technical youth protection settings are important. But they are only a small part of media education and should primarily be seen as support for parents and children. Far more important than technical settings is for parents to accompany their children in an age-appropriate manner and to teach and exemplify a safe and healthy approach to digital media. In our parent brochure "My child and AI - growing up with artificial closeness", we focus on AI chatbots and AI companions in media education. These systems are programmed to build relationships with humans and simulate emotional exchange. The brochure explores the question of whether and how young people can be emotionally influenced by AI bots. It also offers parents concrete support for media education with ten practical tips as well as impulse questions that promote dialog between parents and children. The brochure also contains a glossary with the most important terms relating to AI systems.
