AI Ireland founder Mark Kelly joined Claire Byrne on Newstalk this past Friday to address a phenomenon sweeping the nation: the “Brad-watch” deepfakes.
With Hollywood star Brad Pitt currently in Ireland filming the A24-backed mystery The Riders (directed by Edward Berger), fans have been scouring Dalkey in Dublin, Timoleague in Cork and Lough Tay in Wicklow for a glimpse of the actor. However, social media has been flooded with “candid” shots of Pitt in local chippers and pubs that are, in reality, sophisticated AI fabrications.
While these celebrity sightings are largely lighthearted, they represent a “proof of concept” for a much darker reality. As Mark Kelly emphasised to Claire Byrne, if we can be easily fooled by a Hollywood star in a jumper, we are equally vulnerable to the malicious deepfakes now targeting Irish elections and bank accounts.
What Exactly is a Deepfake in 2026?
A deepfake is synthetic media (images, video or audio) created using Generative Adversarial Networks (GANs) or Diffusion Models. In 2026, these tools have evolved beyond simple face-swapping to include:
- Real-time Video Synthesis: Used in video calls to impersonate colleagues or family members.
- Voice Cloning: Requiring as little as 3 seconds of reference audio (often pulled from a person’s social media) to create a perfect vocal replica.
- Contextual Manipulation: Placing real people in fake environments, such as the recent AI-generated images of Brad Pitt appearing to film at Bullock Harbour in Dublin weeks before he actually arrived.
Why It’s the “Year of the Deepfake” in Ireland
2026 is a pivotal year for Ireland. Between the Presidential Election cycles and high-profile film shoots, the “shadow economy” of AI disinformation has exploded. In late 2025, a deepfake of candidate Catherine Connolly went viral, and a fake “RTÉ News AI” channel was terminated by Facebook in October 2025 for spreading synthetic misinformation.
Advanced Signs: How to Spot a Deepfake
During the Newstalk interview, Mark Kelly shared that while AI is getting better, it still fails at the “edges of physics.” To spot a deepfake in 2026, look for these specific anomalies:
A. The “Turn-the-Head” Test
Most deepfake models are trained on front-facing data. In a video call, ask the person to turn their head to a full profile. You will often see the ear blur, the jawline detach from the neck, or glasses “melt” into the skin.
B. Lighting and Shadows (The Irish Sky Test)
AI often struggles with the physics of natural light. Check if the shadows on the subject’s face match the ambient lighting of the Irish background. Many “Fake Brad” images show him under studio-quality lighting while standing under a characteristically grey, overcast Dublin sky.
C. Biological “Glitches”
- Unnatural Blinking: Humans blink spontaneously every 2–10 seconds. AI faces often stare for unnaturally long periods or blink with a mechanical, rhythmic pattern.
- Teeth Morphing: Watch the teeth when the person speaks. In deepfakes, teeth can often appear as a single white block or subtly change shape mid-sentence.
- Breathing Patterns: Listen for the breath. Synthetic voices often insert breaths at syntactically incorrect moments or lack the natural “micropauses” found in human speech.
The Threat to Irish Businesses and Citizens
The danger extends far beyond celebrity pranks. In the first quarter of 2026, deepfake-driven fraud led to over $200 million in global financial losses.
Financial Scams in the Irish Market
New research by Revolut and Juniper Research (Feb 2026) shows that social media platforms generated €32 million from scam advertisements targeting Irish consumers last year. The average loss per scam is slightly over €1,500. Scammers are increasingly using deepfakes of Irish politicians or “trusted” financial figures to promote fake government investment opportunities.
“Business Voice Compromise”
A growing trend in 2026 is the targeting of Irish SMEs. Scammers use voice cloning to call a junior employee, impersonating the CEO’s voice to authorise an “urgent” wire transfer.
The Legal Landscape: Ireland and the EU AI Act
As Mark Kelly noted, the law is finally beginning to catch up.
- The EU AI Act: Most provisions, including transparency obligations, become enforceable in August 2026. This will require professional “deployers” of AI to clearly label deepfakes.
- CoCo’s Law (Ireland): This remains a powerful tool against the non-consensual sharing of intimate deepfakes, carrying heavy criminal penalties.
- The EU Presidency: Ireland is expected to use its EU Presidency in late 2026 to push for even stricter harmonised laws regarding AI-enabled harms.
Frequently Asked Questions (FAQ)
Q: Why are people making deepfakes of Brad Pitt in Ireland?
A: Primarily for “engagement bait.” High-profile celebrities generate clicks and follows. Once an account has enough followers, it can be sold to scammers or used to spread disinformation.
Q: Can my phone detect a deepfake automatically?
A: Not perfectly. While tools like McAfee’s Deepfake Detector or mobile security apps exist, they are currently in an “arms race” with AI models. Human scepticism is still your best defence.
Q: What is a “Safe Word” and why do I need one?
A: Mark Kelly recommends families establish a “safe word”- a random, unshared phrase like “Lego Teapot.” If you receive an emergency call from a “family member” asking for money, ask for the safe word. If they can’t provide it, it’s likely a voice clone.
Q: Are deepfakes illegal if they are just for parody?
A: Under the EU AI Act, satirical or creative deepfakes must still be disclosed as “artificially generated” unless the disclosure interferes with the “enjoyment” of the work (e.g. in a film). However, using a celebrity’s likeness to promote a commercial product without consent is a violation of personality rights.
Q: What should I do if I am a victim of a deepfake scam?
A: 1. Report to the Platform: Flag the content on X, Facebook or Instagram. 2. Contact your Bank: If money was moved, alert your fraud department immediately. 3. An Garda Síochána: Report the incident to your local station, especially if it involves identity theft or harassment.
Conclusion: The Path Forward for AI Ireland
As we navigate the “Brad-watch” and the upcoming political cycles, awareness is our primary shield. At AI Ireland, we believe in harnessing the power of AI for positive innovation – while remaining vigilant against its misuse.
As Mark Kelly told Claire Byrne, “The technology isn’t going away; our ability to critically evaluate what we see and hear must grow alongside it.”
Would you like AI Ireland to host a “Deepfake Defence” workshop for your organisation? Contact us today to learn more about our 2026 Digital Literacy programmes.
Call to Action
If you’d like to delve deeper into how these trends can reshape your organisation, we would be delighted to discuss them in more detail. Invite Mark Kelly, Founder of AI Ireland, to speak at your next team meeting, conference or strategy session. We can explore practical ways to harness AI responsibly, meet sustainability goals, and navigate the evolving consumer landscape. Let’s work together to ensure Ireland remains at the vanguard of innovation in 2026 – and beyond.
Discover more from AI Ireland
Subscribe to get the latest posts sent to your email.
