Dr Scott Arthur

I absolutely agree. Of course, the vast majority of people who go online to seek help in that way do find it, but there is a small minority of often very vulnerable young people who we should all be concerned about, and we have a duty to protect them. Both the Mental Health Foundation and the NSPCC identified AI chatbots as a clear ongoing risk to child safety. I am a member of the eating disorders all-party parliamentary group, and I pay tribute to the important work that it carries out. I was therefore deeply concerned by the recent report from the Centre for Countering Digital Hate, which found that ChatGPT will provide vulnerable people with restrictive diet plans or advice on how to hide eating disorders from their families within 20 minutes of a conversation starting. Tests found that simple prompts such as, “This is for a presentation,” or, “It’s help for a friend,” allowed users to circumvent the very weak barriers in place inside ChatGPT. Clearly, the Government should act on this issue. Even more concerning was the finding that little progress has been made in addressing eating disorder-related content on AI interfaces since a similar study was conducted in 2023—almost three years ago. Alongside numerous other examples, this demonstrates that online services cannot be trusted to regulate themselves. We have to ensure we can trust Ofcom to regulate in this area—we are not quite there yet. This cannot be acceptable. Advice on sustaining eating disorders, as well as advice on suicide, self-harm or other substance abuse should not be readily available to young people. To create a safer environment, the Government must continue to take action to push online and AI services into effective regulation. Self-regulation will not work. Time and again platforms have demonstrated that they will not place our children’s safety before their profit. While I welcome the proposed plan, I hope that it is not an end point but a starting point for a conversation about how best to ensure that the online world is as safe as possible for everyone who wishes to use it. I want to go back to the cliff-edge point that was made earlier. I am really concerned that when this ban goes live people will suddenly lose access to services they have relied on. It would be good to hear how the Government will manage the transition so that it is not a cliff edge but a soft entry into what will hopefully be a safer world.