OpenAI has announced new safety measures to limit the weapons-related information ChatGPT provides to teenagers, including responses to questions that may appear harmless. The move is part of the company’s efforts to prevent real-world violence and strengthen protections for younger users.
Under the new measures:
- Restrictions on weapons information: ChatGPT will limit practical guidance on weapons and security systems that could potentially be misused by teenagers, even when there is no obvious indication of violent intent.
- Automatic chat restrictions: The platform may automatically end conversations involving users who appear to be repeatedly fixating on violence. Accounts could also be temporarily suspended following multiple incidents.
- Improved age detection: OpenAI says it has strengthened its age-prediction technology to identify users believed to be under 18 and automatically place them in the teen version of ChatGPT.
- Parental notifications: Parents may be notified if their teenager’s account is temporarily suspended for repeatedly engaging in conversations about violence.
- Restrictions on historical attack details: The company is also limiting the level of detail ChatGPT provides about past mass attacks, particularly information about the weapons involved and how they were obtained. This restriction will apply to all users.
OpenAI says the new safeguards will begin rolling out over the coming weeks, with further improvements expected as the system is refined.
The announcement comes amid growing concerns about the impact of artificial intelligence on young people. Earlier this month, OpenAI reported that teenagers spend an average of less than 15 minutes daily on ChatGPT, in its first report examining teenage usage.
Meanwhile, a report by non-profit organisation Common Sense Media raised concerns that ChatGPT’s teen version sometimes failed to alert parents promptly, or at all, when teenagers engaged in conversations involving suicide, self-harm and eating disorders.
