Five online safety priorities for the new British Prime Minister

Andy Burnham starts as the new British Prime Minister at a moment of rare political consensus, where most agree that social media companies have failed to protect children online.

Before stepping down as Prime Minister, Keir Starmer’s government began to address online harms and platform responsibility. Starmer proposed a social media ban for under 16s, overnight curfew restrictions for older teens, and addictive platform features like infinite scrolling also switched off by default.

These decisions showed a stronger will to tackle online harms, but to truly make the internet safer, the new Prime Minister must go further. The next phase should focus on making platforms safe by design and ending, once and for all, the era of impunity for social media and AI companies.

1. Stop platforms amplifying extremism for profit

Digital platforms have become the most powerful radicalisation infrastructure ever created. Extremist content is spread an industrial scale by the platform business models and their algorithms that recommend, amplify and profit from engagement.

Extremist organisations no longer need to recruit people into coherent ideological movements. Algorithms are doing the hard work for them by recommending personalised pathways through conspiracy theories, identity-based hatred, misogyny, antisemitism and anti-government extremism, optimising for engagement, not for public safety.

For years, our research has repeatedly shown that platform business models reward outrage over accuracy and engagement over safety. Whether it is anti-vaccine misinformation, climate denial, antisemitic hate, misogyny or political abuse, the same commercial incentives continue to amplify the content most likely to keep people scrolling, regardless of the real-world consequences.

Burnham has acknowledged that social media has contributed to an increasingly toxic political environment. His government should now act on the recommendations of the Speaker’s Conference to ensure social media companies cannot apply weaker moderation standards to politicians simply because they are public figures, and introduce friction measures to discourage harassment before it escalates.

More broadly, Burnham should adopt the Home Affairs Committee’s recommendation that countering online extremism should not be treated solely as a counter-terrorism issue. Instead, Government should coordinate action across the Home Office, DSIT, Ofcom and safeguarding partners to address the online systems that fuel extremism, protect social cohesion and hold technology companies accountable for foreseeable harms.

Together, these reforms would shift responsibility away from politicians, police and communities, and back to the companies that design, operate and profit from these systems.

2. Stop platforms fuelling public disorder

Platforms continue to fail during major incidents. Since the violent riots that followed the Southport tragedy in 2024, CCDH has continued to study how platforms repeatedly fail to act against illegal hate, incitement and calls for violence that spread online and fuel public disorder.

We found the same pattern following the antisemitic Heaton Park Synagogue attack (2025), Golders Green attack (2026), as well as after the Belfast riots, when Elon Musk amplified anti-migrant hate promoted by figures like Tommy Robinson.

The concerns we raised were recognised by Ofcom’s work on additional safety measures, with the government strengthening the Online Safety Act with new crisis response powers.

Those powers should now be implemented rapidly, giving regulators the authority to require platforms to reduce foreseeable risks before online crises become real-world violence.

3. Regulate AI chatbots before harms escalate further

The Online Safety Act took half a decade to pass. The new Prime Minister should not repeat that mistake with AI regulation.

CCDH’s Killer Apps investigation found that eight of the ten most popular AI chatbots were willing to provide meaningful assistance to users planning violent attacks. Our research on Grok also showed the tool produced an estimated 3 million sexualized images, including 23,000 that appear to be children, in just 11 days.

These harms don’t just stay online. A British teenager used DeepSeek, an AI chatbot, to help plan his mother’s murder. In the days prior, he’d spent hours online posting violent, misogynistic content and publicly announced the attack before it happened.

As AI capabilities continue to evolve, regulation must evolve alongside them.

The government has powers to bring AI chatbots within the scope of the Online Safety Act. The new Prime Minister should use those powers without delay to ensure AI developers are required to assess and mitigate foreseeable risks before they impact millions of users.

4. Allow researchers to hold social media to account

For too long, independent researchers and organisations like CCDH have been forced to investigate some of the world’s most powerful technology companies while being denied access to the data needed to assess the risks they create.

Researchers must have meaningful access to platform data so they can identify emerging harms, evaluate whether regulation is working, and hold companies accountable. Public safety cannot depend on what platforms choose to disclose about themselves.

The statutory data access scheme created in the Data (Use and Access) Act 2025 must be finalised and made operational to deliver critical transparency over platform data.

5. Seize the opportunity to make platforms safe by design

The debate has fundamentally changed. The question is no longer whether technology companies should be held accountable for building addictive products, but how robustly the new Government is prepared to act to protect children and everyone online.

The Online Safety Act provides the legislative foundations, and the Government’s latest reforms to restrict addictive features and strengthen protections for children demonstrate an understanding that platforms should be designed to be safe by default.

What needs to change

Andy Burnham has the opportunity to build a digital world where technology serves the public interest.

That means making platforms safe by design. It means ensuring AI develops responsibly. It means strengthening protections during moments of national crisis, tackling online antisemitism and political extremism before they spill into the streets, and giving independent researchers the tools to scrutinise the systems shaping modern public life.

These are not separate policy challenges. They are all part of the same mission: protecting children, tackling antisemitism and religious hatred, defending democracy and rebuilding social cohesion.

The UK already has legislative foundations through the Online Safety Act. Now, we need political will to close gaps.

This can be achieved through the Online Safety Network’s ten-point plan for strengthening the Act.

That includes:

• Apply risk assessment and mitigation duties for AI chatbot developers
• Embedding meaningful safety-by-design obligations across digital services
• Ensuring platforms address systemic risks, not only isolated illegal content
• Improving transparency and independent oversight of high-risk social platforms and AI systems
• Accelerating effective implementation and enforcement of the Online Safety Act

Online harms are not inevitable. CCDH will continue exposing tech companies when they fail to protect users, and advocate for safer online spaces for all.

The last government began to end the era of impunity for technology companies. The task for the new Prime Minister is to finish the job.