Last week, amid considerable political tension in the UK over a planned pro-Palestine march through London on Armistice Day, an audio clip from the mayor of London, Sadiq Khan, was circulated on the social media platform TikTok.

“I don’t give a flying sh*t about the Remembrance weekend,” Khan said. “What’s important and paramount is the one million-man Palestinian march that takes place on Saturday.”

In another clip, Khan observed: “I know we have Armistice Day on Saturday but why should Londoners cancel the Palestinian march on Saturday? Why don’t they have Remembrance weekend next weekend? What’s happening in Gaza is much bigger than this weekend and it’s current.”

A few weeks ago, a recording circulated in Slovakia, days before that country’s general election, of a conversation between the leader of the Progressive Slovakia party, Michal Šimečka, and a journalist in which the two men dismissed ordinary voters, discussed buying votes from the country’s Roma minority and joked about child abuse imagery. In another recording released the day before the election, Šimečka advocated increasing the price of beer – not, in Slovakia or anywhere else, a policy likely to be popular with ordinary voters. Šimečka lost the election narrowly, to the Smer party, led by Robert Fico and considered friendlier to Russia and Hungary.

Both the Khan and Šimečka recordings were “deepfakes”, counterfeits produced on computers with the assistance of artificial intelligence (AI). Such hoaxes were possible before powerful computers met AI; now, they can be completely indistinguishable from the real thing. With lightning speed, they can be spread everywhere by social media platforms, while traditional media is fretting about whether it should ask if they are true, or fact-checking the sources. The lie is halfway around the world while the truth is getting its boots on.

They can be harmless, of course. People will probably recall the picture of the pope in a puffa jacket. Last summer, a publicity shot of Rishi Sunak pulling a pint was doctored to show him pulling a disgracefully messy pint at the Great British Beer Festival – and it flew online.

Or not harmless. A deepfake of Ukrainian president Volodymyr Zelenskiy calling on his soldiers to lay down their arms was uploaded to a hacked Ukrainian news website. During the Turkish election in May, the opponents of under-pressure incumbent Recep Tayyip Erdoğan found clips of themselves being endorsed by the Kurdistan Workers’ party, the PKK. The PKK is designated as a terrorist organisation in Turkey, the US and the UK.

On and on it goes. As you would expect, US president Joe Biden is a regular target. One of many was a deepfake video of Biden, produced with AI technology, announcing a national draft to recruit US soldiers to send to Ukraine. Can you imagine what’s going to happen next year in the presidential election, where – according to recent polling by the Pew Research Center – a third of adults under 30 get their news from TikTok?

Here, we have two elections for sure next year – the local and European polls in early June – and possibly a general election. And, like much of the rest of the world, we are utterly unprepared to defend our democratic processes from the threat posed by the new technology in the hands of unscrupulous participants in the contest, or malign outside actors.

[ University College Cork study finds ‘deepfakes’ from Ukraine war undermining trust in conflict footage ]

It’s not hard to imagine the sort of thing that could emerge as a deep fake. A video of Micheál Martin indicating he would enter coalition with Sinn Féin. Of Mary Lou McDonald admitting she really admires King Charles. Of Leo Varadkar saying he actually, totally hates the Greens.

“I have no doubt that a believable deepfake will emerge at a critical stage in an Irish election,” says Fianna Fáil senator Malcolm Byrne, one of the few Irish politicians to have thought about all this and who has been agitating for the Government to take action. “By the time it is denied, it will have been shared thousands of times on social media and, even with the denial, some will still choose to believe it.”

I think there are three things that are needed – and needed urgently.

Firstly, we need a comprehensive legal and regulatory framework, including strong investigation and enforcement powers to hold bad actors and tech companies to account. The good news is that there is legislation already on the statute books and a regulatory body in the new Electoral Commission. The bad news is that the Government has not commenced parts 4 and 5 of the legislation, which would give the commission the power to monitor and regulate election campaigning online, including paid advertising, and also enforce its decisions on internet and social media companies. The Government says it is dealing with EU objections, but the European Commission says it is awaiting to hear from the Irish Government about how it proposes to overcome their concerns. I get little sense that this will all be up and running before next year’s elections, in part because of the objections of the tech companies to intrusive regulation.

And that’s the second thing we need – a real determination by the tech companies to protect the integrity of our democratic and electoral systems. This needs to go beyond tinkering with the algorithms and instead involve properly policing the content that they facilitate and disseminate – even when it is expensive and disruptive for them. Frankly, our democracy is more important than their profits.

The third thing we need is for the voting public to educate themselves about all this and to learn how to assess all the information flooding towards them with a discerning eye. This is our democracy. We all need to learn how to protect it.

QOSHE - It’s not hard to imagine the fun bad actors might have using AI to disrupt Irish politics - Pat Leahy
menu_open
Columnists Actual . Favourites . Archive
We use cookies to provide some features and experiences in QOSHE

More information  .  Close
Aa Aa Aa
- A +

It’s not hard to imagine the fun bad actors might have using AI to disrupt Irish politics

12 28
18.11.2023

Last week, amid considerable political tension in the UK over a planned pro-Palestine march through London on Armistice Day, an audio clip from the mayor of London, Sadiq Khan, was circulated on the social media platform TikTok.

“I don’t give a flying sh*t about the Remembrance weekend,” Khan said. “What’s important and paramount is the one million-man Palestinian march that takes place on Saturday.”

In another clip, Khan observed: “I know we have Armistice Day on Saturday but why should Londoners cancel the Palestinian march on Saturday? Why don’t they have Remembrance weekend next weekend? What’s happening in Gaza is much bigger than this weekend and it’s current.”

A few weeks ago, a recording circulated in Slovakia, days before that country’s general election, of a conversation between the leader of the Progressive Slovakia party, Michal Šimečka, and a journalist in which the two men dismissed ordinary voters, discussed buying votes from the country’s Roma minority and joked about child abuse imagery. In another recording released the day before the election, Šimečka advocated increasing the price of beer – not, in Slovakia or anywhere else, a policy likely to be popular with ordinary voters. Šimečka lost the election narrowly, to the Smer party, led by Robert Fico and considered friendlier to Russia and Hungary.

Both the Khan and Šimečka recordings were “deepfakes”, counterfeits produced on........

© The Irish Times


Get it on Google Play