The company said these tighter restrictions apply even if the question asked “appears harmless.”
OpenAI is clamping down on “weapons-related information” teens can get from ChatGPT, even if their question “appears harmless,” the company announced in its latest post. OpenAI noted that it already has policies in place that prevent ChatGPT from helping users carry out violent acts or teens getting their hands on weapons. To further limit those scenarios, OpenAI said it’s going to “restrict practical information about weapons and security systems that could be misused to facilitate violence, even when there’s no sign of violent intent” with its teen users.
The AI giant noted that teens will still be able to use ChatGPT when it comes to preventing or de-escalating violence, seeking support and connecting with a trusted adult. Not just with teens, OpenAI is strengthening policy boundaries for all of its users, particularly when asked about “reconstructions of past mass attacks, including details about the weapons and equipment used and how they were obtained, and detailed comparisons on scale of harm that could amplify an attacker’s notoriety.”
However, these restrictions will work alongside the parental controls that OpenAI put into place last year, which sends notifications to parents if their linked teen accounts have violated ChatGPT policies related to violent threats or acts of violence. These additional safeguards come after OpenAI was sued in a wrongful death lawsuit last year and when the company’s CEO, Sam Altman, publicly apologized for not alerting police to alarming chat logs in relation to the Tumbler Ridge shooting earlier this year. Looking ahead, OpenAI said it will publish its first Community Safety Report in early 2027, which will be followed by updates every six months.
