Media & information · Platforms and algorithms · United Kingdom · 2024–2026
Feeds now pick much of the news people see. Who checks how they choose?
Online intermediaries carried news to 73% of UK adults in the past month (Ofcom 2026). Their choices shape what people see. Evidence of amplification is stronger than evidence of changed votes; safety regulation does not amount to a requirement for balanced news.
Evidence last reviewed 3 October 2026
Three things the evidence establishes
1 · They are where news is found
73%
of UK adults saw news through an online intermediary in the past month. Ofcom 2026; not directly comparable with 2025.
Which services, and how much attention?
This includes social media, video, search, aggregators, messaging and AI tools. Intermediaries took 37% of 72 daily news minutes. Meta reached 51% of adults with news. Separately, YouTube reached 94% of adult internet users for 51 minutes a day in 2025; that measures all use, not news alone.
2 · They are built for engagement
Ofcom’s own guidance warns that recommender systems “may amplify hateful content if they are optimised for user engagement”. Across 2.7 million US political posts, language about opponents was the strongest predictor of being shared. In Ofcom’s June 2025 tracker, scrolling a feed was the most common way people met potential online harms (36% of encounters); misinformation was the most commonly encountered harm (41% in the past four weeks).
Ofcom illegal harms guidance; Rathje et al. 2021; Ofcom Online Nation 2025
3 · Their news choices are not regulated
UK law regulates feeds for harm: illegal content and content harmful to children. Ownership, plurality and impartiality rules cover TV, radio and newspapers but not the ranking that decides which news a platform user sees. Duties to protect news publishers’ content on the ten largest platforms were still out to consultation in September 2026.
What UK law requires of recommendation feeds
| Duty | What it covers | Status |
|---|---|---|
| Illegal content risk assessment and safety duties | All in-scope services must assess how their design, including algorithms, could spread illegal content, and act on the risks. | In force; enforceable since 17 March 2025 |
| Children’s safety duties | Services likely to be used by children must assess and reduce children’s exposure to harmful content, including content an algorithm pushes repeatedly. | In force from summer 2025 (risk assessments due 24 July 2025) |
| Transparency reports, including on algorithms | Categorised services must publish annual reports with information about the algorithms they use and their effect on users. | Legislated; applies to categorised services |
| News publisher content, journalistic content, content of democratic importance, user control | Extra duties for the ten Category 1 platforms, including protecting recognised news publishers’ content and letting users control what they see. | In law; Ofcom still consulting on how they apply (September 2026) |
| Children’s data used to recommend content | Under the ICO Children’s Code, services are responsible for recommendations they make to children using their personal data. | In force since September 2021 (ICO) |
| Plurality or balance of the news a feed shows | No UK rule. Media ownership and impartiality rules apply to TV, radio and newspapers, not to platform ranking. | Not regulated |
Category 1 platforms — the top tier, defined partly by using a content recommender system — are Facebook, Instagram, Pinterest, Reddit, Roblox, Snapchat, TikTok, WhatsApp, X, YouTube (Ofcom register, last updated 28 September 2026). The Act treats misinformation as in scope only where it is illegal or harmful to children, or where a Category 1 platform’s own terms prohibit it.
Ofcom’s guidance puts the risk in its own words: “Our evidence, for example, indicates that if not properly tested and deployed, content recommendation systems may amplify hateful content if they are optimised for user engagement.” The ICO applies a different test to children’s feeds: “If you are using children’s personal data to automatically recommend content to them based on their past usage/browsing history then you have a responsibility for the recommendations you make. This applies even if the content itself is user generated…”
No confirmed Ofcom enforcement action specifically targeting a platform’s recommender system — as opposed to risk-assessment paperwork, age assurance, or generated content — has been found in this research as of 2026-09-07. This is stated as "not found," not "does not exist": it is exactly the kind of live regulatory fact that can change without notice, and should be re-checked directly against Ofcom’s enforcement pages before being relied on.
What is contested: do feeds change minds?
The best causal evidence comes from the US. In a randomised experiment during the 2020 election, switching Facebook and Instagram users from the ranked feed to a chronological one substantially changed what they saw and how long they stayed — but produced no measurable change in polarisation or political knowledge over three months (Guess et al., 2023). That is a genuine complication for the strongest claims about algorithms and politics; it is not proof feeds have no effect, because three months is short and the study could not test effects on the whole information system. A 2024 systematic review of 112 studies found that the field’s disagreement about “filter bubbles” tracks research method rather than converging on one answer (Hartmann et al.).
Evidence: Well established, contested effect size
Moral-emotional contagion
Why does outraged or moralised content spread further than neutral content?
How it works and the studies behind it
Messages containing moral-emotional language (words that are both about morality and charged with emotion) are retweeted at a higher rate than otherwise similar messages — in one large study, each additional moral-emotional word raised a tweet’s diffusion by 20%. The effect is stronger within a political in-group’s own network than across the divide, which is one documented pathway by which morally charged content can deepen rather than bridge political separation.
Brady, Wills, Jost, Tucker & Van Bavel, "Emotion shapes the diffusion of moralized content in social networks," PNAS 114(28), 2017 (n=563,312 tweets on three US political topics) — explicitly correlational; the authors themselves call for controlled experiments to test causation on attitudes and behaviour, not just retweet counts.
Evidence: Well established, contested effect size
Out-group animosity as an engagement driver
What single content feature best predicts whether a political post gets shared?
How it works and the studies behind it
Across 2.7 million Facebook and Twitter posts from US news outlets and members of Congress, language about the political out-group was the strongest predictor of shares and retweets found — posts about opponents were shared about twice as often as posts about one’s own side, and each out-group word raised the odds of a share by 67%. This out-group effect was several times stronger than either negative-emotion language or the moral-emotional language measured in other studies of this kind.
Rathje, Van Bavel & van der Linden, "Out-group animosity drives engagement on social media," PNAS 118(26), 2021 (n=2,730,215 posts). The authors state plainly that the design is correlational and that "it is unclear how algorithmic choices... might contribute to the amplification of out-group animosity, since social media companies are not transparent about how their algorithms work" — i.e., this paper documents what content people reward with engagement, not what a ranking algorithm then does with that signal.
Evidence: Weak
Engagement-optimised ranking as a business-model choice
Why would a platform’s algorithm favour emotionally arousing content at all?
How it works and the studies behind it
Feed-ranking algorithms are built to maximise predicted engagement — the metric platforms’ advertising business depends on — not to maximise accuracy, civility or representativeness. Academic reviewers describe content algorithms as an "attentional filter" that pre-selects the moral/emotional content people are already disposed to notice, "designed to increase engagement and profit for the platforms." A former Facebook product manager testified to the same underlying logic under oath: the company "programs its algorithms to maximize profits, which means it decides which speakers are heard and which are not," and Facebook’s own 2017 decision to weight "angry" reactions as five times more valuable than a "like" in its ranking formula is documented by investigative reporting on internal company documents.
Brady, Crockett & Van Bavel, "The MAD Model of Moral Contagion," Perspectives on Psychological Science 15(4), 2020 (peer-reviewed theoretical review, not new experimental data on the algorithm itself). Frances Haugen, Written Testimony to the US House Energy & Commerce Subcommittee, 1 Dec 2021 (sworn testimony from a named former insider; not independently verified against the underlying internal documents by this research pass). Merrill & Oremus, "Five points for anger, one for a ‘like’," Washington Post, 26 Oct 2021 (investigative reporting on leaked internal Facebook documents; documents themselves not directly examined). Rated Weak because no source isolates the ranking algorithm’s own causal contribution from the entangled effects of human sharing behaviour, moderation policy and account/monetisation design — the mechanism is well-argued and consistent with insider testimony but not independently measured end-to-end.
Evidence: Contested — credible evidence points both ways
Filter bubbles and the "backfire" evidence
Does algorithmic personalisation trap people in one-sided information, and does removing it fix polarisation?
How it works and the studies behind it
The "filter bubble" thesis (Eli Pariser, 2011) argues personalised ranking narrows what people see to content that confirms their existing views, deepening division. The empirical record on this is genuinely split, and splits along method: studies measuring actual browsing/exposure data have found that individual choice about what to click narrows a person’s information diet more than Facebook’s ranking algorithm does, and that social/search channels can even increase exposure to opposing views compared with direct browsing. Separately, a large randomised experiment that switched real Facebook and Instagram users from the algorithmic feed to a plain chronological one during the 2020 US election found it substantially changed what they saw and how long they spent on the platform — but did not produce a measurable change in political polarisation, political knowledge, or belief in false claims over the three-month study. A 2024/2025 systematic review of 112 studies found the field’s disagreement tracks research method (network-based computational studies tend to confirm echo chambers; content-exposure and survey studies tend to reject the strong version of the thesis), not a converging answer.
Pariser, "The Filter Bubble," 2011 (the originating thesis; a book, not an empirical study). Bakshy, Messing & Adamic, "Exposure to ideologically diverse news and opinion on Facebook," Science 348(6239), 2015 (n=10.1m users; Facebook-employee authors). Flaxman, Goel & Rao, "Filter Bubbles, Echo Chambers, and Online News Consumption," Public Opinion Quarterly 80(S1), 2016 (n=50,000 browsing histories). Guess, Malhotra, Pan, Barberá et al., "How do social media feed algorithms affect attitudes and behavior in an election campaign?", Science 381(6656), 2023 (n=44,764 across Facebook/Instagram; Meta-funded, several co-authors are Meta employees — disclosed here alongside the finding). Hartmann, Pohlmann, Wang & Berendt, systematic review of 112 studies, J. Computational Social Science, 2025 / arXiv:2407.06631, 2024 — finds the literature’s disagreement is methodological, not yet resolved either way.
Evidence: Well established, contested effect size
Social-feedback reward loop for outrage expression
Does getting likes for angry posts make people post angrier content again?
How it works and the studies behind it
In preregistered Twitter studies and behavioural experiments, receiving positive social feedback (likes, shares) for an outrage expression increased the likelihood a user would express outrage again — a pattern consistent with reinforcement learning. Users also matched their outrage expression to their network’s normative level of outrage, an effect that was strongest where the reward signal (feedback) was the main available cue and weaker where the network norm was already highly visible.
Brady, McLoughlin, Doan & Crockett, "How social learning amplifies moral outrage expression in online social networks," Science Advances 7(33), 2021 (two observational Twitter studies, n=7,331 users / 12.7m tweets, plus two preregistered behavioural experiments, total N=240). The authors state the observational studies "cannot draw causal inferences" on their own (self-selection into extreme networks is a plausible alternative explanation) and note the behavioural experiments, while causally identified, use a simulated rather than real platform environment.
Southport, 2024: what is documented, and what is disputed
After three girls were killed at a dance class in Southport on 29 July 2024, a false name and false claims about the attacker spread online within hours, while reporting restrictions kept broadcasters from naming the suspect. The case shows exposure and amplification clearly. It does not show, on its own, that a ranking algorithm caused the disorder that followed — that question is kept separate below.
Chronology
- 29 Jul 2024A 17-year-old attacked a Taylor Swift-themed children’s dance class in Southport with a knife. First 999 call 11:47am; suspect arrested at the scene within ten minutes.
- 29 Jul 2024 (same day)The false name "Ali Al-Shakati" began circulating on social media, originating from the site Channel3Now, alongside false claims the attacker was a Muslim asylum seeker who arrived by small boat.
- 30 Jul 2024Two of the three girls attacked died (a third died in hospital the same week). By 3pm the false name had over 30,000 mentions on X from over 18,000 unique accounts. Merseyside Police publicly stated the name was incorrect.
- 30 Jul 2024, eveningA vigil outside Southport Mosque was followed by violent disorder: bricks and rocks thrown at the mosque, a police vehicle set alight, a shop looted; 39 officers injured.
- 31 Jul 2024 onwardDisorder spread to towns and cities across England and to Belfast; over 1,000 arrests followed nationally in the following weeks.
- 1 Aug 2024Liverpool Crown Court lifted reporting restrictions and the suspect was formally named as Axel Muganwa Rudakubana — three days after the attack — the judge citing the risk that continued anonymity would let misinformation keep spreading in a vacuum.
- 22 Oct 2024Ofcom CEO Dame Melanie Dawes wrote to the Secretary of State assessing how illegal content spread after the attack — the fullest primary regulatory document on the case.
- 23 Jan 2025Rudakubana sentenced to life imprisonment (minimum term 52 years) after pleading guilty to all sixteen charges.
- 3 Mar 2025The Law Commission opened a consultation on contempt-of-court law, prompted specifically by the Southport case, on whether more could be published after arrest to counter misinformation.
- May 2025HMICFRS (the police inspectorate) published its Tranche 2 review of the policing response, finding "no conclusive or compelling evidence" the disorder was "deliberately premeditated and co-ordinated by any specific group or network" — directly qualifying the stronger framing in Ofcom’s October letter.
- 6 Aug 2025Amnesty International published an analysis of X’s design and recommender-system choices in relation to the disorder, based on X’s own published recommender source code.
What was measured (not additive)
30,000+ mentions, 18,000+ unique accounts
The false name, on X only, within roughly 24 hours (by 3pm, 30 Jul 2024)
Institute for Strategic Dialogue, "From rumours to riots"27 million+ impressions
Posts stating or speculating the attacker was Muslim, a migrant, a refugee or a foreign national, on X, within the first day or two
Marc Owen Jones, independent academic analysis (corroborated by Amnesty International)580 million+ views
One named account’s (Tommy Robinson’s) total X activity over the two weeks following the attack, not all of it Southport-specific
Amnesty International, "How X’s design and policies led to Southport-linked racist violence"260 million views, 4,379 replies promoting violence
Six named X accounts tracked by CCDH, combined, in the week following the attack
Center for Countering Digital Hate, "Fuelling Hate"
Each figure above is its own organisation’s measurement, over its own time window, on its own platform — they measure different things and are not additive. A widely-circulated "155 million impressions / 1.7 billion reach" aggregate figure could not be traced to any primary source in this research and is deliberately excluded.
Three questions, three levels of evidence
Documented
The false claim spread, and reached a large audience
The false name and false identity claims reached tens of thousands of mentions within a day, and individual posts reached millions of views within days — see the metrics above, each independently sourced.
Documented, but disputed
That spread contributed to the riots
Ofcom’s own letter states illegal content "appears to have contributed to the significant violent disorder." HMICFRS’s statutory review, seven months later, found "no conclusive or compelling evidence" of coordinated premeditation and pointed instead to deprivation, distrust of police and political disaffection — directly qualifying the stronger reading of Ofcom’s letter. Both are authoritative, both are dated, and they do not agree.
Source: Ofcom (22 Oct 2024) vs HMICFRS Tranche 2 review (May 2025)
Not established
Who originated the false claim, and why
Channel3Now published the false name first; Pakistani police arrested an individual linked to the site on 20 August 2024, then dropped the case six days later for lack of evidence he was the originator. The BBC found no evidence for separately-circulated claims of Russian state involvement. No source establishes who wrote the false claim or why.
Source: Wikipedia (Channel3Now), cross-checked against contemporaneous press reporting
“Accounts (including some with over 100,000 followers) falsely stated that the attacker was a Muslim asylum seeker... Posts about the Southport incident and subsequent events from high-profile accounts reached millions of users, demonstrating the role that virality and algorithmic recommendations can play in driving divisive narratives in a crisis period.” — Ofcom, 22 October 2024
What this case cannot be used to conclude
- That "the algorithm" specifically — as opposed to a platform’s moderation policy, its verification/monetisation design, or individual users’ own sharing choices — was the cause of the false name’s spread. Ofcom names "virality and algorithmic recommendations" as a factor; Amnesty’s report analyses X’s recommender code specifically; neither isolates the algorithm’s contribution from these entangled design choices.
- That online misinformation was the cause, or even the primary cause, of the riots. HMICFRS’s statutory review and the Children’s Commissioner explicitly pushed back on that "prevailing narrative," citing deprivation, distrust of police and political disaffection as more complex underlying drivers.
- Who specifically originated the false claim, or why. The one investigation into this (Pakistani police, against a Channel3Now-linked individual) was dropped for lack of evidence.
- A precise reach total. The individual figures above measure different platforms, accounts and time windows and are not additive — and a widely-circulated aggregate figure could not be traced to any primary source and is excluded.
What would change it
Options include independent researcher access to platform data, finalising and enforcing the Category 1 duties on news and transparency, and extending plurality reviews to online distribution — each with its own trade-off against expression and feasibility. They are compared in Transparency and plurality options.
Sources and methods (10)
- Ofcom, News consumption in the UK 2026 — research findings, Figures 1, 8 and 9; adult data tables, Table 359 (released 15 September 2026) — news reach and attention through online intermediaries (2026 survey; not directly comparable with 2025)
- Ofcom, Online Nation 2025 (Online Experiences Tracker, June 2025; Ipsos iris, May 2025) (released 10 December 2025) — platform use, where potential harms are met (36% via feeds), misinformation encounters (41%)
- Online Safety Act 2023 — the statute
- DSIT and DCMS, Online Safety Act: explainer (released updated 24 April 2025) — how the Act treats algorithms and misinformation; implementation dates
- Ofcom, Register of categorised services (Online Safety Act 2023) (released published 30 June 2026, last updated 28 September 2026) — Category 1 list and duties in consultation
- Ofcom, Illegal harms: risk assessment guidance and risk profiles (December 2024) — Ofcom on recommender systems as a risk factor
- ICO, Age appropriate design: a code of practice for online services — Standard 12, Profiling — children’s data and recommendations
- Guess et al., “How do social media feed algorithms affect attitudes and behavior in an election campaign?”, Science 381 (2023) — randomised experiment on feed ranking
- Rathje, Van Bavel and van der Linden, “Out-group animosity drives engagement on social media”, PNAS 118 (2021) — what content drives sharing
- Ofcom, letter from Dame Melanie Dawes to the Secretary of State on the Southport disorder (22 October 2024) — Ofcom’s assessment of the Southport case
Method notes
- Reach, amplification and influence are treated as separate steps, each with its own evidence. Most causal studies of feeds are from the US; UK-specific causal evidence is scarce.
- Southport figures come from different organisations, platforms and time windows and are not added together.