OpenAI makes ChatGPT less 'human' for teens in new safety update

OpenAI has announced a series of new safety features for ChatGPT aimed at under-18 users. Teen users will be able to disable the human voice response, reducing the AI's ability to appear human. Regular reminders will encourage young people to take breaks from using ChatGPT, reminding them it's artificial intelligence and that it can wait.

OpenAI insists these changes are not a response to specific issues but address growing concerns about people mistaking AI tools for sentient beings. The company is following a similar teen account model used by major social networks. Teen accounts will become the default for users who identify as teenagers when signing up, with others identified by the company as under 18.

Under-13s will not be able to use the product. Study Mode, which helps students with schoolwork without giving away answers, will now be the default setting for teens during designated hours. Quiet times will allow teens to switch off ChatGPT. A majority of ChatGPT users, according to OpenAI, use the tool for learning. A recent report by the Policy Exchange think tank raises concerns about the potential for AI-enhanced grade enhancement in UK universities, with 94% of surveyed students admitting to using AI in assessments.

OpenAI will now flag eating disorder-themed prompts to parents, reviewing each alert and sending it within an hour. The company aims to address moments when offline support is crucial. No bespoke AI regulation exists in the UK, though the former government considered restricting AI chatbots for under-18s. While Anthropic's AI tool Claude remains adult-only, Google's Gemini AI tool initially prohibited children from using it.

Experts warn that excessive AI usage could lead to 'ultra-processed minds', likening it to concerns about ultra-processed foods.