Fake AI-generated ‘good news’ stories are flooding social media, with fabricated tales of politicians performing heroic acts designed to exploit positive emotions and drive mass engagement, fact-checkers have warned.
One of the most widely shared examples features a photograph of Nigel Farage, dressed in a suit, crouching on a step while gently patting a dog wrapped in a blue cape. The accompanying post claims the Reform UK leader and his partner Laure Ferrari “heroically rescued” 47 dogs after buying an entire shelter that was on the verge of closure. “Forty-seven lives saved. No spotlight. Just heart,” it reads. The post attracted around 4,000 comments, including one that read: “Many people will vote for a person who shows compassion to animals.” But the photograph is not real, and the story is entirely fabricated.
The post is one of roughly 100 similar items that investigators at the charity Full Fact have uncovered across social media. Together, the posts have generated more than 380,000 total reactions, suggesting large numbers of users are being taken in by false narratives that the charity says are being “churned out at an industrial scale”.
Industrial‑scale fakes targeting politicians across the spectrum
Other examples identified by Full Fact include a false story that Mr Farage had donated millions to open homeless support centres across Kent, another that claimed he had saved abandoned baby twins, and a third that said he had given up his first‑class plane seat to a military veteran. An AI‑edited image showed Mr Farage alongside a sick child, falsely claiming he had visited the child to grant a dying wish. Deepfake videos have also circulated, including one of Mr Farage playing Minecraft and destroying Rishi Sunak’s virtual home.
The former prime minister Rishi Sunak has also been targeted. A fake story claimed he was recovering in hospital, while more than 100 deepfake video advertisements impersonating him were promoted on Facebook, some reaching up to 400,000 people, often pushing a fraudulent “Quantum AI” investment platform. Another deepfake showed Mr Sunak instructing schoolchildren on how to play Fortnite. Restore Britain leader and MP Rupert Lowe, and Reform’s home affairs spokesman Zia Yusuf, are also subjects of bogus posts. A fake post claimed Mr Lowe had called for a ban on sign language teaching – the work of a satirical account – while a fabricated “explosive clash” between Mr Lowe and BBC presenter Laura Kuenssberg was also circulated. Mr Yusuf has been targeted by coordinated bot networks aiming to discredit him, with some of the attacks estimated to have reached nearly half a million views.
Full Fact found that while many of the pages had names suggesting a UK focus – ‘Britain Awakens’, ‘British Affairs Review’ and ‘Political Brief UK’ – most were managed by accounts appearing to be based in Vietnam, sometimes alongside accounts from the United States or Hong Kong. The charity stressed there was no suggestion that any of the pages were associated with Reform UK or Mr Farage. A spokesperson for Reform UK told Full Fact the claims in the posts were false and that the party was not affiliated with any of the pages.
The ‘weaponisation of empathy’
According to experts, the tactic represents a deliberate shift in how AI‑generated disinformation is being deployed. Instead of relying on outrage or fear, these campaigns are designed to “weaponise empathy” – using heartwarming narratives to bypass critical thinking and trigger strong emotional reactions that social media algorithms reward.
“What’s striking about these posts is that they are offering fake ‘good news’ stories, leveraging empathy and positivity rather than outrage to drive engagement,” said Steve Nowottny, editor of Full Fact. “The fact the posts we saw have had hundreds of thousands of reactions shows that many are falling for these false narratives.” He warned that accessible AI tools are making it “easier than ever for fictional slop to be churned out on an industrial scale – potentially allowing content creators to profit as a result”.
Sam Stockwell, senior research associate at the Centre for Emerging Technology and Security at the Alan Turing Institute, described the posts as “weaponising empathy rather than hate”. He explained: “Overseas content creators are now using AI to craft heartwarming fakes, knowing that social media algorithms prioritise content which trigger strong emotive reactions.”
Professor Martin Innes, co‑director of Cardiff University’s Crime and Security Research Institute, said he is seeing content creators “churning out” misleading posts “at scale”, often generating revenue off the back of it. “Where these kinds of visual disinformation and distortion used to take a reasonable amount of input, that is not the case anymore,” he said. “And as for the emotional register they are pitched in, that is just a way of trying to secure views in a noisy and cluttered attention order.”
Researchers at the Alan Turing Institute have noted that while AI‑enabled misinformation may not have yet decisively swung election results, there is growing concern about the “persistent erosion of confidence in what is real and what is fake across our online spaces”. Voters may confuse legitimate political content with AI‑generated material, undermining public trust in democratic discourse.
Meta, the parent company of Facebook, has said it takes action against such content. A spokesperson said: “We have clear community standards that prohibit harmful misinformation and inauthentic behaviour and we have removed these accounts for violating our policies.” Meta is also implementing an “AI info” label for realistic AI‑generated images, videos or audio that could mislead viewers, with such content being downranked, and it requires advertisers to disclose AI‑generated or heavily edited content in political and social issue ads. Despite these measures, Full Fact found that new pages are being created almost daily, and the platforms appear unable to halt the spread entirely.
Mr Nowottny called for sustained action rather than a reactive approach. “So while it’s welcome that Meta has taken action against the accounts we identified, we can’t simply play misinformation whack‑a‑mole – we need sustained action from social media companies to ensure that AI content is identified more smartly.” The rapid proliferation and sophistication of AI‑generated political misinformation is outpacing existing regulatory frameworks, including the UK’s Online Safety Act, which requires online services to assess the risk of fraud and remove illegal content but faces the challenge of an ever‑accelerating stream of synthetic material.
