Cork Experts: Dodgy AI deepfakes are delivering dissent

Following the recent presidential election, there is growing concern among Cork cyber experts about the potential dangers posed by deepfake AI.

A few days before the election, what appeared to be a convincing deepfake AI-generated video, which mimicked an RTÉ news report, falsely showed Catherine Connolly announcing the withdrawal of her candidacy from the race.

Deepfakes consist of images, videos, or audio that have been edited or generated using AI tools or audio-video editing software.

This is the latest incident highlighting the threat that deepfake AI poses to politics and democracy, not just in Ireland but across the world.

Speaking to The Echo about the deepfake AI video that appeared during the presidential election, Stephen Treacy, lecturer in digital transformation at University College Cork, said: “My own reaction to it is that it wasn’t surprising.

“This is something that has been a long time coming. Misinformation and disinformation campaigns are nothing new.

“But the tools with which they are being deployed, now quicker than ever, are really a cause of concern,” said Mr Treacy.

“We are getting to the stage now, where technology is almost blurring reality and fiction.

“So, it is getting harder and harder to distill what truth means and whether truth has a role in online society anymore.”

AI in 2023 Slovakian Election

Days before the September 2023 parliamentary elections in Slovakia, AI-generated audio clips were spread online targeting Michal Šimečka, leader of the pro-European Progressive Slovakia party.

The deepfakes falsely depicted Mr Šimečka discussing election rigging and increasing the price of beer.

Mr Šimečka went on to lose the election, highlighting concerns that AI-generated clips had cost him votes.

“From the Catherine Connolly perspective, there are bad actors out there that are intent on using these technologies to sow dissent and cause confusion, especially around electoral timelines,” said Mr Treacy.

“We have seen this in Slovakia, over in the United States, Pakistan, and Argentina. There are loads of real contemporary examples out there.

“But the worrying thing, certainly from my perspective, is that they just keep getting better, and it is getting very difficult to identify what is false.”

Stephen O’Keeffe, director of cyber and privacy at PwC in Cork, told The Echo that from his perspective seeing a deepfake used to the extent that it was in the Irish presidential election campaign did not come as a major surprise.

“That is because deepfake technology has been around for a little bit of time, and the quality has been improving. We have seen that elections are a common use for disinformation.

“I would say that the example of Catherine Connolly in the presidential election in Ireland was a high quality one though, and it shows an example of what is possible these days.”

Rise in deepfakes

According to a European Parliament report, a projected eight million deepfakes will be shared in 2025, considerably up from the figure of 500,000 in 2023.

Europol, the law enforcement agency of the European Union, has estimated that 90% of online content may be generated synthetically by 2026, as deepfakes spread rapidly through social media platforms, messaging apps, and video sharing platforms, further blurring the line between reality and fiction.

Mr Treacy said: “I think it has got to the stage now where everyone has to question whatever they are being presented with online and maybe triangulate that with a couple of different sources to make sure it is actually legit.”

In response to political deepfakes, the EU’s AI Act, passed in 2024 with key provisions taking effect in 2025, requires AI-generated content like deepfakes to be clearly labelled. The EU AI act also sets transparency rules for platforms.

Meanwhile, the EU Digital Services Act (DSA) is a wide-ranging EU regulation that creates a safer and more transparent digital space for users.

Organisations and tech platforms are working to raise public awareness regarding the existence and the threats of deepfakes, encouraging people to report suspected fake content.

Mr O’Keeffe said there will be an onus on individuals and organisations to be even more aware in terms of preventing a deepfake attack.

“Companies are probably going to have to invest in detection tools. I think it is another tool in the arsenal that security programs are going to have to look at,” Mr O’Keeffe said.

“Ultimately, what is required is that you have incident response playbooks ready to go. That is because you have to prepare for a scenario where you do fall foul of something like this.

“You need to be prepared to react, to response and to minimise the damage. That is really important.”

With regards to what he feels needs to be done to address deepfake AI attacks, from a regulatory and technological response, Mr Treacy said it is important that organisations be proactive rather than reactive.

“It is extremely dangerous. We are in new territory,” Mr Treacy said. “There is a huge onus on the platforms to mitigate this and be responsible for their information.

“Are we seeing much of an appetite from them to really clamp down on these videos?

“It is definitely something that from a legislation point of view, needs to be looked at more thoroughly.”

Protection

Although the methods used to tackle cyber-crime are evolving, so are the ways in which hackers use deepfake AI.

Mr Treacy said people should not be afraid to question or verify content they are consuming online.

“The scary thing is that in the space of two years, we are now looking at these videos where you have to do a second take to say, ‘is this real or isn’t it?’.

“Your eyes are now capable of being deceived by these technologies and that is something that we are seeing as a new phenomenon for society.

“I don’t think that there is this concept of truth anymore in the online world that we live in.

“Truth is outdated. Truth can be manipulated. Truth can be weaponised,” Mr Treacy said.

“I would really urge hesitation in sharing, pushing, or proliferating a narrative that you’re not overly convinced is legit.

“Double check sources. Take a look at other news outlets, news organisations, journalists,” he added.

“Just because you see it in one video, if it’s a bit of a big deal and if it’s trying to cause a bit of chaos, chances are it’s going to be picked up elsewhere.

“They will have done a bit of due diligence. They will have made sure that this actually is legit.

“That way you can place a little bit more stock in it, as opposed to coming across a random video online.”