OpenAI curbs ChatGPT’s spoken replies for teenage users

OpenAI has announced that its AI chatbot will

OpenAI has announced that its AI chatbot will no longer engage in spoken, voice-based conversations with users under the age of 18. The change, which takes effect immediately for new accounts and rolls out over the coming weeks for existing ones, is part of a broader effort to promote safer habits among younger audiences. The company stressed that the move was not triggered by any single incident involving minors treating the system as a living companion.

Instead, the adjustment reflects a growing concern among child-development experts and regulators about how synthetic voices can blur the line between a tool and a person. OpenAI’s own research has found that voice interactions—especially those with emotional inflections—are more likely to foster a sense of social presence. Teenagers, in particular, are at a developmental stage where such cues can feel more real and influential, potentially affecting how they perceive trust and authority.

The new policy means that while teens can still type chat messages freely, they will no longer hear ChatGPT speak back to them in a natural, human-sounding tone. The voice mode was introduced earlier this year and had been popular among younger users for homework help and casual conversation. OpenAI said the restriction applies to all ChatGPT interfaces, including mobile apps and web browsers, and that parental controls remain available to manage overall usage.

For adult users, nothing changes—voice interactions will…

For adult users, nothing changes—voice interactions will continue as before. However, the company is also adding a new transparency label for audio outputs that clearly identifies them as machine-generated. This is meant to reinforce the idea that ChatGPT is not a person, even if it sounds like one. The update arrives amid broader industry scrutiny over AI companionship features, with other firms facing similar questions about emotional attachment among minors.

OpenAI has not disclosed specific metrics on how many teens used voice mode, but the decision signals a cautious approach ahead of potential regulatory guidelines. Child safety advocates have welcomed the move, though some argue that written interactions could also carry risks if they become too personal. The company says it will continue to review its safeguards as the technology evolves, and it encourages parents to discuss the limits of AI with their children. The voice ban for teens is expected to be permanent, unless future research shows a way to make such interactions measurably safer.