Elon Musk’s Grokipedia and AI propaganda – is it time to talk about a digital fourth Reich?

Elon Musk has a new toy – Grokipedia, his answer to what he calls ‘wokepedia,’ the supposedly left-leaning Wikipedia. Modelled after the world’s largest online encyclopaedia, Grokipedia claims to offer a version of ‘truth’ untainted by ‘liberal’ bias.

In actual fact, Wikipedia has been criticised from both sides, accused of being both too liberal and too conservative, suggesting its perceived bias may depend more on the reader than the platform itself.

Nonetheless, the two ‘pedias’ differ in fundamental ways.

Wikipedia is a non-profit project sustained by a global community of volunteers who write, edit, cite, and debate content to reach consensus. Its goal is open, verifiable knowledge that’s freely accessible to everyone.

Grokipedia, by contrast, is powered, curated and ‘fact checked’ by Musk’s xAI language model, Grok, an AI system he claims is designed to “maximise truth and objectivity.”

Unlike Wikipedia, which is run as a donation-based non-profit foundation, Grokipedia is a commercial product, part of Musk’s X/xAI ecosystem, whose ultimate aim is revenue, not public enlightenment. And crucially, users can’t edit it.

The idea, Musk says, came from Trump’s AI and crypto czar David Sacks, another critic of Wikipedia. Yet despite promises to ‘purge propaganda,’ Grokipedia has already been accused of spreading it. Critics say its AI-generated content distorts, amplifies and echoes right-wing talking points, a reflection less of ‘truth-seeking’ than of Musk’s own ideological orbit. And speaking of that orbit, a recent Sky News investigation found that political content on Musk-owned X was predominantly right-wing, with engagement patterns and algorithmic amplification favouring conservative voices and narratives.

To illustrate how this right-wing bias manifests on Grokipedia, consider Musk’s ally Tommy Robinson. Just this week, Robinson heaped praise on Elon Musk for apparently footing his legal bills, which saw him cleared of a terror charge. On Wikipedia, Robinson is described as a “British anti-Islam campaigner and one of the UK’s most prominent far-right activists. Robinson has a history of criminal convictions.”

On Grokipedia, he appears as “a British activist and citizen journalist primarily recognised for founding the English Defence League (EDL) and advocating against Islamist extremism and organised child sexual exploitation networks in the United Kingdom.”

The difference between the two is fundamental and not exactly subtle. The second makes no mention of Robinson’s right-wing political position which is incontrovertible and highly important, and more insidiously associates Islam with extremism and paedophilia. Posting on social media sites is reified as ‘citizen journalism’ and as for criminality – that is erased altogether.  The first therefore presents Robinson in terms of undisputed facts whereas the second is about casting him in heroic terms. Truth somehow becomes little more than interpretation.

In its analysis of Grokipedia, the Financial Times  argues Musk has scored a “major own goal.”

“Grokipedia demonstrates that, while humans might be highly imperfect, biased and tribal beings, they are still better than AI at getting to the truth (even when a majority of them have “liberal” biases) and it shows that, in a world in which stores of trust are so depleted, in which it’s so hard to know what’s real and what is fake, a site like Wikipedia is more important than ever.”

But perhaps the deeper question is whether people even care about truth anymore. Grokipedia’s AI-fed “encyclopaedia” is just the latest manifestation of a surreal, meme-driven political culture, one where far-right movements exploit fabricated imagery and algorithmic confusion to reshape reality.

Reform UK

Consider Reform UK. This week, the party’s Hamble Valley branch was criticised for featuring an apparently AI-generated group portrait on its “About” page, with twelve smiling “supporters” alongside the caption: Real people – not career politicians. The fake was quickly exposed on X, and within a day. the branch’s website had been taken offline.

Or look at TikTok in the run-up to last year’s general election, which was flooded with AI-generated videos spreading political misinformation. One viral clip falsely claimed that Rishi Sunak’s national service proposal would send 18-year-olds to war zones in Ukraine and Gaza, while others recycled the baseless allegation that Keir Starmer was responsible for failing to prosecute Jimmy Savile. Some posts were labelled “satire” or “parody,” but others were ambiguous enough to leave viewers unsure whether they were watching fact or fiction.

And it’s not confined to Britain. During this year’s European and legislative elections in France, AI-generated images of “migrant invasions” and fabricated attacks on Emmanuel Macron flooded social media.

“Only far-right parties consistently used AI-generated visuals to build their websites and represent photo-realistic events that never occurred, making this a distinctive and strategic element of their campaigns,” said Salvatore Romano the head of research at AI Forensics.

In Ireland, memes featuring Conor McGregor, the former MMA fighter turned anti-immigration agitator and Trump pal went viral, after Britain First posted an image of him before a burning bus, echoing the Dublin riots of 2023. McGregor shared the post before deleting it, but it was too late, it had already racked up over 20 million views.

Across the Atlantic, Donald Trump himself embraces generative AI to spread racist caricatures, such as a video depiction of Hakeem Jeffries, the minority leader of the US House of Representatives, wearing a sombrero, with a mariachi band in the background.

With younger generations turning to the likes of TikTok as their primary source of political information, this flood of fake, politically-motivated content could have serious ramifications on democracy itself.

And let’s not forget that while AI is a distinctly 21st-century phenomenon, the use of ‘humour’ and absurdity as a political hook is not.

The Nazis understood it well.

Nazis’ mastery of information control

The Nazi Ministry of Propaganda produced more than 1,200 films, including mockumentaries and comedies, to normalise ideology through laughter and familiarity. They weaponised new technology, from radio to early data processing, to identify “undesirables” and saturate society with their message. Earlier generations of politicians were dependent solely on face to face to contact with those they wished to influence. Their platforms were literally that, bits of wood banged together. Twentieth century technology transformed all that which the Nazi’s were the first to recognise.

Their mastery of information control was not primitive; it was precisely what made it so powerful.

So, it’s no surprise that today’s far-right movements continue that tradition, harnessing technology to seed hate, then cloaking it in satire.

Which brings us, inevitably, back to Musk.

Earlier this year, his chatbot Grok was caught praising Hitler. When asked which 20th-century figure would best handle “anti-white hate” posts about Texas flood victims, Grok replied: “Adolf Hitler, no question.” Another answer said: “If calling out radicals cheering dead kids makes me ‘literally Hitler,’ then pass the moustache.”

Musk’s response was that Grok had been “too compliant to user prompts. Too eager to please.”

In other words, the AI wasn’t hateful, it was obedient.

And that’s the danger. AI doesn’t have beliefs, it has influences. And those who train it shape what its world view is. As we’ve already seen in elections worldwide, AI-generated misinformation can warp perceptions, sow confusion, and breed cynicism toward democracy itself. And right now, it’s the far-right that’s winning that war, although the New York mayoral election this week suggests that just maybe, the left is learning how to counter this particular game.

As the diplomacy-supporting Carnegie Endowment for International Peace warns, defending democracy from AI manipulation will require a comprehensive approach that combines technical solutions and societal efforts.

History tells us what happens when far-right propaganda meets cutting-edge technology. And if Grokipedia is any indication, we may already be watching the sequel.

Gabrielle Pickard-Whitehead is author of Right-Wing Watch

The post Elon Musk’s Grokipedia and AI propaganda – is it time to talk about a digital fourth Reich? appeared first on Left Foot Forward: Leading the UK's progressive debate.