PRESIDENTIAL candidate Catherine Connolly has become the target of an AI-generated deepfake video that has gone viral across social media.
The video falsely claims that Connolly has withdrawn from the race and that the election scheduled for this Friday has been cancelled.
Styled as an official RTÉ News broadcast, the video uses artificially generated versions of well-known broadcaster Sharon Ní Bheoláin and political correspondent Paul Cunningham to deliver the news.
Connolly then appears to confirm her own withdrawal, speaking in a voice that closely mimics her own at what is made to look like a real campaign event.
A digitally created supporter is seen shouting, “No, Catherine!” as the scene unfolds, followed by further fake analysis that names rival candidate Heather Humphreys as the automatic successor to the presidency.
These videos had already reached nearly 200,000 views on Facebook before they were taken down.
According to Meta, Facebook's parent company, deepfakes are typically removed within seconds, but in this case, the video remained live for over 12 hours.
Previous videos from the same fake account, “RTÉ News AI”, remained online for days and featured deepfaked versions of other RTÉ journalists falsely claiming that over half of all presidential ballots had been spoilt.
The account mimicked RTÉ’s branding and font design to appear authentic and deceive the casual viewer.
Dr Barry Scannell, a member of the state’s artificial intelligence advisory council, warned that this form of AI misuse poses a critical threat to democracy.
“The level of sophistication of this deepfake is shocking,” he said, according to the Irish Independent.
“People can and will take these at face value without digging further.”
The false reports continued to circulate even as Connolly appeared alongside Heather Humphreys on RTÉ’s Prime Time debate.
The Connolly case is part of a wider trend in the misuse of artificial intelligence to create convincing but entirely fake videos.
While companies like OpenAI and Google Gemini maintain strict policies against generating politically deceptive material, other platforms with fewer restrictions make it easy to produce realistic deepfakes.
These tools are increasingly being used for scams and fraud as well.
