Labour’s Pro-BBC Censorship Push is Not Based on Evidence

Labour’s online censorship plan to force social media companies to promote the BBC and other ‘trustworthy’ news outlets and demote the rest is not based on evidence, according to a new report by David Rose for the Free Speech Union. David writes about his report in today’s Telegraph; here’s an excerpt.

In the past year, ministers have said they want to impose broadcast-style rules of accuracy and balance on social media platforms before elections, to compel them to push ‘trustworthy’ news into every citizen’s feed. Everything else will be demoted.

In other words, barely noticed by opposition politicians or the media, the Government is rolling the pitch for the most restrictive regime to control public speech since the reign of Charles I.

They justify each of their measures with the same freedom-chilling argument: that misinformation and disinformation, pumped out to innocent users by malevolent algorithms engineered to monetise extremism, division and hatred, now threaten democracy itself. The narrative has been repeated so often that it has acquired the status of fact. But the evidence suggests it is largely false.

Consider the detail of what is proposed. Lucy Powell, Labour’s Deputy Leader, wants platforms to be treated like broadcasters during elections because of the impact of what she says are “social media feeds decided by opaque algorithms, where falsehoods, deepfakes and co-ordinated mis and disinformation can spread at alarming speed”.

Culture Secretary Lisa Nandy’s Green Paper Watch this Space proposes laws to give “regulated public service media” prominence in social media feeds, so that sources such as the BBC are “seen and heard in the fierce battle against mis and disinformation”.

Meanwhile, the media regulator Ofcom is imposing new rules requiring major platforms to identify and restrict the circulation of content suspected of being illegal during “extraordinary events” such as the riots that followed the Southport murders in 2024.

Behind all this sits an “evidence review” which was published earlier this year and commissioned by the Government Office for Science (GO-Science). It claims online misinformation has become “a systemic national risk”, blames algorithms and platform business models for its spread, and calls for “system-level” intervention, including the compulsory integration of “quality signals” into the ranking of content.

In March, Steve Reed, the then communities secretary, told the Commons that voters had been driven into “online echo chambers exacerbated by malevolent algorithms”.

Yet, as I reveal in a report to be published this week by the Free Speech Union, research evidence published by prestigious academic journals suggests otherwise.

In 2024, Nature issued an authoritative review by a team led from the University of Michigan. Its conclusions were the reverse of the Government’s premise. It found that exposure to misinformation and extremist content is low on average and heavily concentrated among a small fringe of users who actively seek it out, rather than coming across it by accident thanks to malign algorithms.

Visits to 490 websites deemed “untrustworthy” by self-appointed fact-checking organisations made up less than 6% of visits to news sites. Meanwhile, “fake news” accounted for some 0.15% of the average American’s daily media diet. According to the paper, exaggerating the threat was creating a risk of disproportionate measures that would threaten free speech.

In 2023, Science published the result of an experiment in which algorithms dictating the feeds of tens of thousands of Facebook and Instagram users were switched off. Content the algorithms ‘thought’ they wanted to see was replaced with feeds based on the chronological order of when content was posted.

After three months, there was no measurable change in their political attitudes, and it turned out the algorithms served less untrustworthy content than a chronological feed, not more. The same journal later found that misinformation about COVID-19 vaccines played a very small role in persuading people not to be vaccinated, and that accurate stories published by legacy media outlets about people who fell ill after having jabs were far more influential.

None of this appears in the GO-Science review. Astonishingly, its assertion that platform algorithms push divisive content because they are “exclusively” driven by profit and engagement rests on a single citation: a 2025 paper from Universitas Muhammadiyah Buton, a small private institution in the Indonesian province of Southeast Sulawesi, ranked 470th of 562 universities in Indonesia and 12,117th of 14,131 in the world. …

Even more troubling is that the word “misinformation” is being stretched to cover dissent, not just untruths. A label that purports to denote falsehood is being applied to legitimate opinion, and to accurate statements that officialdom finds unhelpful. For example, in 2020, the suggestion that the Covid virus escaped from a laboratory in Wuhan was widely branded a conspiracy theory. Since then, the CIA and FBI have concluded a lab leak is the likeliest explanation, and several high-profile scientists support this view.

The GO-Science review says misinformation requiring Government attention includes dangerous “media narratives” that “decouple” Net Zero from climate science and challenge the validity of electric vehicles and heat pumps.

The review’s catalogue of falsehoods also lists “anti-LGBTQ+ narratives”, “‘gender ideology’ claims targeting schools” and “transgender athlete controversies”.

Worth reading in full. And read David’s full report for the FSU here.

Stop Press: Watch FSU External Affairs Officer Connie Shaw discuss the new report on GB News.