AI chatbots could be ‘restricted for children’ under new social media ban

Sir Keir Starmer is considering an accelerated ban on social media for under-16s, including restrictions on AI chatbots.

The Prime Minister wrote on his Substack yesterday that he wants to ‘tighten’ the Online Safety Act to ensure AI companies remain ‘firmly within its scope’.

He said: ‘It is crucial that our rules keep pace with the rapid pace of technological change.

“This builds on recent steps we’ve taken to ban nudification apps and criminalize taking intimate images without consent.”

Concerns about chatbots – computer models trained with vast amounts of text to predict the next word in a sentence – were echoed by Liz Kendall.

The technology secretary told Times Radio this morning: ‘If it’s just one-to-one with AI chatbots, I, like the Prime Minister, am really concerned about the impact this has on children and young people.

“And I would say that we are taking steps so that all illegal content shared by AI chatbots will be stopped for everyone – including adults.”

Kendall said AI technology companies will be held accountable by media regulator Ofcom, just as social media companies do now.

“That’s why the Prime Minister said, ‘You don’t get a free pass here; you are responsible for your technology,'” she added.

If AI companies are forced to comply with the Online Safety Act, they must stop their bots from generating illegal content related to terrorism, child sexual abuse, harassment and hate crimes or risk being fined.

It comes after the government clashed with X over its built-in chatbot, Grok, which captured explicit images of real people and children.

The UK’s Online Safety Act already requires social media networks and adult websites, such as porn, to age check users.

Yet Subway has revealed how AI bots encourage young people to consider suicide, while children can chat with bots pretending to be Jeffrey Epstein.

Meanwhile, child safety and mental health experts are increasingly raising the alarm about children viewing chatbots as friends and therapists.

What could a social media ban entail?

The government last month launched a consultation on banning social media for teenagers after the Lords backed a rebel amendment calling for it.

The world’s first social media ban on young people came into effect in Australia in December, requiring people to be over 16 to use most platforms.

Platforms are expected to use age estimation technologies for this, such as facial and voice analysis or examining activity patterns.

But Starmer’s take on the ban may go a step further than that, according to his Substack post.

Starmer said he wants to set a minimum age for social media “within months” and limit features that “keep them hooked to screens” such as autoplay and endless feeds.

He added that he plans to limit VPN access for children, referring to the technology that creates a secure connection between your computer and a server in another country.

This would be “to make it more difficult for children to circumvent the age limits of services or certain features,” Starmer wrote.

These will be considered in the consultation on the ban on social media.

Social media ban ‘like sticking blunder over a pothole’

Child safety campaigns Subway spoke with mixed feelings about Starmer and Kendall’s announcement.

Katie Freeman-Tayler, head of policy and research at Internet Matters, said two-thirds of children use chatbots.

She added: ‘While we welcome the government’s announcement to protect people from illegal content on AI chatbots, it is important that they also hold AI chatbot providers accountable in protecting children from content that is legal but harmful to them, such as suicide and disordered eating.

Bertie Aspinall, co-founder of parental safety tool Safetymode, said the government is working on the assumption that “no amount of social media on phones will make smartphones safe.”

“While there is bad content on social media and on AI chatbots, it’s like pasting bluster over a pothole,” Aspinall said.

“It may seem like it’s working, but it’s not really working.”

Instead, Aspinall suggested, officials should ask phone makers to bring in addictive features.

Sophie Stocks, vice-president of education and wellbeing at online safety group Smoothwall, says children need to be taught how to use technology responsibly.

“While schools are already doing extraordinary work to protect students in their care, a lack of time and resources means that it is unfortunately difficult to stay ahead of the curve when it comes to online harms, especially AI, and many early warning signs can go unnoticed or misread,” she added.

“Today’s announcements will help us move towards a more agile, proactive system fit for the threats we face.”

Contact our news team by emailing webnews@metro.co.uk.

For more stories like this, view our news page.

MORE: Keir Starmer warns Britain must ‘step up and move faster’ on defense spending

MORE: Which Of These 4 New Personality Types Do You Fall Into?

MORE: Mock The Week drags Strictly Tess Daly and Claudia Winkleman into Keir Starmer drama