OpenAI limits weapons information for teens using ChatGPT

KUALA LUMPUR, October 31 - OpenAI is introducing stricter controls on weapons-related data accessible through ChatGPT for users under 18, complemented by enhanced measures designed to identify and deter potential violence. The company announced these additional guidelines in a statement released yesterday, emphasizing that the new rules would restrict practical information about weapons and security systems that could be misused, even if a teenager's queries seem innocent or show no signs of violent intent.

The company clarified that its existing policies already prohibit aiding anyone in planning or executing violent acts and preventing minors from obtaining or using weapons. OpenAI emphasized that these safeguards are being broadened to restrict practical information about weapons and security systems that could facilitate violence, even when no immediate intent of harm is evident.

In addition to these measures, OpenAI announced stricter limitations on all users regarding information about past mass attacks. This includes detailed descriptions of the weapons and equipment employed, how they were acquired, and comparisons that could sensationalize the perpetrators' notoriety. The company is also developing automated systems to terminate conversations when users exhibit signs of obsession with violence or a serious intent to harm others. Persistent instances of such behavior may lead to temporary suspensions of their accounts.

OpenAI intends to adopt a more cautious approach when past conversations reveal warning signs of potential violence, even if subsequent questions appear harmless in isolation. The company disclosed that it had referred over 200 cases of potential threats of violence to law enforcement authorities during the third quarter of 2026, asserting that such referrals had played a crucial role in disrupting violent plots.

The company stated that its age-prediction technology will continue to evaluate users' behavior to pinpoint those potentially under 18, applying extra protections accordingly. Adults who are mistakenly identified as minors will have the option to verify their age to eliminate these restrictions.

Parents whose accounts are linked to their teenagers' ChatGPT accounts may already receive notifications when their children's accounts are disabled for violating policies on violence. OpenAI is also working on alerts for repeated instances of violent discussions that result in account suspensions.

Separately, OpenAI announced its intention to publish its first Community Safety Report in early 2027, covering the second half of 2026. The report, updated every six months, will explain how the company detects, investigates, and responds to potentially harmful activities. The report will feature anonymized examples of violence-related risks, changes to safety measures, and relevant data on their effectiveness.

OpenAI assured that the new safeguards will be rolled out over the coming weeks as the company continues to refine its systems to detect threats and prevent misuse.