skip to main content

AI-generated deepfake videos target presidential election

The AI-generated videos featured deepfake versions of Catherine Connolly and RTÉ News presenter Sharon Ní Bheoláin
The AI-generated videos featured deepfake versions of Catherine Connolly and RTÉ News presenter Sharon Ní Bheoláin

In the last 24 hours, Meta has removed AI-generated deepfake videos spreading false claims about the Presidential Election. Similar videos remained on YouTube as of Wednesday morning but have since been removed.

In response to the videos, An Coimisiún Toghcháin has urged voters to treat online content with caution. Just two days out from the election, experts say this amounts to disinformation and that the sophistication of the videos could easily fool voters.


A number of AI-generated videos spreading false information about the presidential election have been circulating on social media just days before voters go to the polls.

The videos appeared on Meta’s Facebook platform and Google’s YouTube, using deepfake versions of RTÉ News bulletins to mimic authentic coverage. As of Wednesday morning, the videos had been removed from the platforms.

One video uploaded to Facebook earlier this week, under an account called 'RTÉ News AI', falsely claimed Independent candidate Catherine Connolly had withdrawn from the race and that Friday’s election had been cancelled.

It featured a deepfake version of RTÉ News presenter Sharon Ní Bheoláin in studio saying: "In the last few minutes at a Catherine Connolly campaign event, Catherine Connolly has confirmed her withdrawal from the presidential race."

It then throws to an AI-generated package from Political Correspondent Paul Cunningham, in which a deepfake version of Catherine Connolly says: "It is with great regret that I announced the withdrawal of my candidacy and the ending of my campaign."


An AI-generated video showing a fake RTÉ News report claiming Catherine Connolly has withdrawn from the campaign.


Other videos followed a similar format. One involved a deepfake version of RTÉ presenter Claire Byrne announcing early results from the election. "Shockingly, 56% of the ballots coming in are spoilt ballots."

Another featured RTÉ’s Vivienne Traynor introducing a fabricated report in which the Chair of the National Women’s Council of Ireland said Ms Connolly "belongs in a nursing home rather than Áras an Uachtaráin".

Experts say such content amounts to "disinformation" and is designed to confuse voters about candidates or outcomes.

Liz Carolan, author of The Briefing newsletter on democracy and technology, said this is "electoral process disinformation, which there is absolutely no space for in online discourse and which platforms have committed to making sure doesn't disseminate".

"Any quick eyeballs on this page would have shown you that it's not satire...We know that motivations for something like this could be to mess with the process."

She added: "The timing of this is no accident. It's only a few days before the election. These things are harder and harder to detect when we don't have good transparency for political content."

CATHERINE CONNOLLY AI
The videos featured an AI-generated version of presidential candidate Catherine Connolly

Yesterday, Meta removed the page hosting the AI-generated videos, but not before they were shared and viewed thousands of times. Some comments posted under the videos appeared to believe the content, while others did not believe it was authentic.

As of Wednesday morning, the YouTube channel sharing the videos remained active. Some of the videos on the page had been on the platform for over a month and viewed hundreds of times. The page has since been removed.

In a statement to RTÉ, a YouTube spokesperson said: "We terminated the channel for violating our Community Guidelines, which strictly prohibit channel impersonation. Our teams are working around the clock to safeguard this election, and we're ready to take action on content that violates our policies."

Meta has been contacted for comment.

Ciarán O’Connor of the Institute for Strategic Dialogue said these videos have become a lot more persuasive.

The clip of Ms Connolly withdrawing from the race is "highly convincing," he told RTÉ.

"It depicts real people, in a context we’re used to seeing them and speaking about a topic which many would expect them to be speaking about."

From a technical perspective, he said that "bar some glitches and out-of-sync audio versus visuals, it is quite advanced".

"It requires close examination and attention to discern whether it is genAI (generative AI) or not. But when published and shared on social media platforms where people consume news and information so rapidly, many people may not take the time to assess it, instead believing it to be true and perhaps themselves sharing this disinformation online."

PAUL CUNNINGHAM AI
A deepfake of Political Correspondent Paul Cunningham also featured.

Mr O’Connor said the group behind the channel has been producing and disseminating numerous generative AI videos on the election and other topics for weeks. Another account on X posting similar videos appeared to have been removed in September, he said.

"They were viewed and shared thousands of times," he said. "We do know they have used genAI to promote false information, so it does seem they were intent on misleading public opinion in regards to the election."

He also noted differences between platforms in terms of labelling: "On some, like YouTube, they carried an information label stating they were 'synthetic’ media, while with others, like Facebook, this was not the case."

Dr Barry Scannell, AI Law & Policy lead at William Fry and a member of the Government’s Artificial Intelligence Advisory Council, wrote on LinkedIn about the false video claiming Catherine Connolly withdrew from the race.

"How many will actually check? This video has already been seen by thousands of people."

"This looked so real I had to check other news sites to see were they reporting on it before I thought to check the Facebook page it came from," he added.

"This is the dangerous kind of AI deepfake that harms the democratic process."

An Coimisiún Toghcháin urged voters to be cautious of such content and to ask themselves certain questions when considering information.

"How do I know if this is true? Who is telling me this? Why are they telling me this? Can I consider alternative sources to check factual accuracy?"