“We have tiptoed around this for too long”: Britain takes on information attacks at the UN

Burnham tells the UN that Britain will build a centre against information attacks

Andy Burnham gave his first speech to the UN General Assembly as Britain’s prime minister on 22 September and used it to announce a new body, the National Centre for Information Defence. The centre is to detect information attacks on the country, establish where they come from and disrupt them, and it will bring together the intelligence agencies, government departments, law enforcement and social media companies. Emily Thornberry, who chairs the House of Commons foreign affairs committee, said the centre was the main recommendation of the committee’s March report on how foreign states manipulate information to undermine democracies.

Burnham said the Kremlin spends about £1.3 billion a year on manipulating information. Russian entities, he said, have used bots, fake websites and forged newspaper articles, and have faked the branding of 28 British organisations, among them universities and the BBC. British authorities hold evidence that Russia attempted to interfere in the 2019 general election. He expects AI to make the threats worse, and he named the people he considers most exposed to influence online: those struggling with the cost of living and young people out of work, education and training. “We have tiptoed around this for too long,” he said, adding that Britain had given too much ground to people spreading a corrosive story about life in the country.

The centre will be run at first from the Office for the Prime Minister and Cabinet. Burnham also placed part of the response outside security: his government will try to create opportunities and make life more affordable, because wide inequalities, he said, leave societies at greater risk. Abroad, Britain plans to share what it learns with countries facing elections of their own, as it already has in Moldova, where it helped protect a vote from Russian interference.

From jihadist posters to a fake TikTok fighter: how AI entered Mali’s information war

The Timbuktu Institute, a research centre in Dakar that studies jihadism in the Sahel, has mapped how the armed actors in Mali’s conflict use generative AI: JNIM, a jihadist alliance linked to al-Qaeda; the FLA, a Tuareg separatist movement allied with JNIM since April despite their ideological differences; and Russia’s Africa Corps paramilitaries, who fight alongside the military government. The allies’ coordinated attacks on 25 April killed defence minister Sadio Camara and took the northern town of Kidal.

JNIM’s own material is the most basic of the three. After Kidal its channels put out a propaganda poster carrying the marks of an AI image, smooth surfaces, blended objects, odd details. JNIM lags behind Islamic State in this field, and the institute is more worried about what the tools would let it do next. Machine translation could turn one recruitment message into versions in about 15 regional languages. Each version could be written for a different audience: land and herding rights for young Fulani herders, dignity and a place in society for young men on the edges of the cities. Audio and video could carry the same messages to people who cannot read the group’s written propaganda. Bakary Sambe, the institute’s director, expects recruitment to stop working en masse and start working person by person.

Accounts supporting the FLA go further and put characters on camera. The TikTok account @azzghim, opened in July 2024, is fronted by a young Tuareg fighter with a dark mark above one eye who condemns Africa Corps for violence against people in northern and central Mali; a TikTok data tool cited in the study counted 601,000 followers in July. His bio points to 27 July 2024, the day a three-day battle at Tinzaouaten, on the Algerian border, ended in defeat for Malian forces and their Russian partners. The researchers list what gives the videos away: the pixels of a raised finger vanish into his hair, his curls stay still while a stalk of grain moves in the wind, and a flag in the background jumps where a clip was cut although his voice runs on. No single sign proves AI, the researchers write; together, the anomalies match recognised indicators of manipulated content. Other pro-FLA accounts repost his videos, and two connected accounts look AI-run to the researchers as well, pointing to a network rather than one creator.

Africa Corps runs the most developed operations, which the institute calls a state-level use of AI. Researchers count 19 Russian disinformation campaigns across Mali, Burkina Faso and Niger since 2018, and AI has changed what those campaigns can produce, in Sambe’s words:

The campaign the study documents has approached five million views. It has shown Malian forces defeating the rebels, and after the loss of Kidal it presented Africa Corps in a favourable light and claimed that France and Ukraine armed the attackers. The three countries have meanwhile tightened control over domestic media and cut off international broadcasters whose reporting on the fighting the authorities object to. The researchers also warn that AI will not stay in the information space: it can help identify targets for drone strikes, and it lets armed groups do more with the forces they already have.

Fake ballot videos from two German state elections were pushed to English-speaking audiences

Germany’s domestic intelligence service, the BfV, warned on 16 September that a Russian campaign against Chancellor Friedrich Merz has been growing since the end of June. In its account, given to the business daily Handelsblatt, the campaign spreads fake videos under the names of real news outlets to discredit politicians who take positions against Russia’s interests and to wear down trust in elections. The service named Matryoshka, an operation that copies news brands, and Marc Henrichmann, who oversees the intelligence services for the Bundestag, links the videos to Storm-1516, a network German authorities attribute to Russian military intelligence.

Berlin and Mecklenburg-Western Pomerania voted four days later, and fraud claims spread over the election weekend. One clip appeared to show a man in Berlin cutting up postal ballots marked for the AfD, the anti-immigration party Alternative for Germany. Alina Lipp, a pro-Russian blogger under EU sanctions, shared it, and her post passed 630,000 views on X. The election-monitoring site Wahlrecht.de listed what gave the packages away: no polling cards, unsealed envelopes, papers folded the wrong way.

In Mecklenburg-Western Pomerania, clips that researchers attribute to Storm-1516 claimed that postal ballots in three constituencies left the AfD out. Christian Boden, the state’s election chief, told the press agency dpa that the ballots shown were forgeries, copied from official sample ballots with the AfD row deleted.

A single post on election day passed three million views on X. Much of the reach came from English-language accounts, which reposted the clips with captions presenting the vote as proof that Germany’s establishment rigs its elections, and several of those posts passed a million views each.

The same trick ran before the February 2025 federal election, with a fake Leipzig ballot missing the AfD. Brian Liston of Recorded Future’s Insikt Group, which tracks these operations, found that clip was pushed mainly by English-language accounts in the UK and the US, and that most of its views came from outside Germany. In his reading, the aim was to project the idea of a rigged German vote abroad rather than to convince Germans. He also pointed to the business behind such campaigns: the Social Design Agency, one Russian contractor running them, has to show the Russian government that its work draws attention, because reach targets are written into its contracts.