AI has not invented fraud. People have always lied, impersonated others and exploited trust for money. What AI changes is the speed, cost, realism and scale at which deception can now be produced. A fraudster can gather fragments of a person’s life from social media, company websites and data breaches, LinkedIn profiles, public posts or old photographs.

Generative AI can then turn those fragments into a convincing email, a fake profile, a cloned voice, a synthetic photograph, a realistic video or a chatbot that can keep a conversation going. The scammer no longer needs to be a skilled writer, actor, designer, psychologist and hacker all at once as AI can assist with each of those roles.

We need your consent to load this rte-player contentWe use rte-player to manage extra content that can set cookies on your device and collect data about your activity. Please review their details and accept them to load the content.Manage Preferences

From RTÉ Radio 1’s Today with David McCullagh, Niamh Davenport from Banking and Payments Federation Ireland on how AI scammers are targeting over 50s

What does this look like in practice? Well, imagine for a moment getting a phone call from your daughter. She is crying. She sounds terrified. A stranger takes the phone and says she has been kidnapped. You have minutes to transfer money.

Similarly, imagine joining a video call at work. The faces of authority look familiar. The voices sound right, talking about recent events or projects with confidence. Then a request comes through that is urgent but plausible: approve a payment, share a document, move quickly. That is why the age of AI-enabled fraud should worry us. In both cases, the technology does not have to be flawless. It just has to be convincing enough for long enough.

A useful way to understand the new fraud landscape is to think of five things being manufactured: fake people, fake voices, fake authority, fake messages and fake relationships. Fake people are created through synthetic identity fraud. This can involve AI-generated faces, manipulated documents, plausible biographies, fake social media histories and professional-looking profiles. A fake LinkedIn account can now have a polished photograph, a fluent posting history, a believable employment record and a network of connections.

From RTÉ Radio 1’s Drivetime, Anthropic calls for pause of AI development

This already has organisational consequences. US authorities have warned that North Korean IT workers have used stolen and fabricated identities to obtain remote employment with Western companies. The US Department of Justice said such schemes involved North Korean individuals fraudulently obtaining work with US companies as remote IT workers using stolen and fake identities. In some cases, these workers generated revenue for the North Korean regime while gaining access to company systems.

Fake voices are even more intimate. Voice cloning uses AI-generated audio to mimic a real person’s voice. The most disturbing examples are distress scams, where a victim receives a call from what sounds like a child, partner or family member in danger.

These scams work because they target one of our oldest trust signals: the familiar voice. We are not trained to hear someone we love crying and ask whether the sound is synthetic. We react emotionally before we analyse rationally. These scams do not need to fool someone for a week; it may only need to create panic for two minutes. The European Parliamentary Research Service has warned that AI-powered voice cloning has become one of the most convincing and difficult-to-detect forms of social engineering.

From RTÉ Brainstorm, all you need to know about voice spoofing and audio deepfakes

Fake authority targets trust inside organisations. The most striking example remains the Hong Kong deepfake video-call scam involving engineering firm Arup. In 2024, an Arup employee was deceived into transferring around €24 million after joining what appeared to be a video call with senior company officers. It was later confirmed that fake voices and images had been used and that the incident had been reported to police.

This was not simply a badly worded email pretending to come from a boss, but a staged performance of corporate authority. The employee was placed inside a simulated organisational setting: faces, voices, hierarchy, urgency and instruction. The scam worked by making the fraudulent request feel socially and professionally legitimate.

Fake messages are also becoming harder to dismiss. Traditional phishing emails were often full of spelling mistakes, strange phrasing and vague greetings. Those errors gave people useful warning signs. AI removes many of them. A fraudulent email can now be fluent, polite, context-aware and professionally formatted. It can refer to an actual project, copy a manager’s tone, arrive during a busy work period or appear to come from someone with authority.

From RTÉ Radio 1’s This Week, the Central Bank’s Head of Enforcement, Market Abuse and Oversight Peter Gallagher on how they’re now targeting illegal financial promotions on social media to protect Irish consumers

The FBI has warned that chat generators can quickly create official-sounding emails mimicking CEOs or other company officials, including emails that contain phishing links or instructions to wire funds. It also states that businesses reported losses of more than $30 million in 2025 to business email compromise scams involving AI. A phishing email does not need to defeat a firewall if it can defeat a tired employee at 4.55pm on a Friday.

Finally, fake relationships are becoming easier to scale. Romance scams, investment scams and confidence scams depend on time, emotional pressure and trust. Generative AI can help sustain conversations, produce affectionate replies, translate across languages, adapt to a victim’s responses and create supporting materials such as photographs, documents or testimonials.

This is not about panic, but about recognising that the trust infrastructure of everyday life is changing

A romance scam once required a human criminal to maintain long conversations with each target. AI can help generate the scripts, the emotional cues and the daily responses. A fake investment opportunity can be promoted by a cloned celebrity, supported by synthetic testimonials, explained by a polished chatbot and reinforced by professional-looking documents.

AI is turning deception into a production line: personal data is the raw material, generative AI is the machinery and human trust is the target. The outputs are money, passwords, identity documents, access to systems and emotional control. Trust now needs a second step. Families need to talk openly about voice scams before they happen and create code words for emergency calls. Employees need permission to pause suspicious requests, even when they appear to come from senior people.

From RTÉ Six One News, fraudsters are using AI-generated images to target people online

Organisations should require second-channel confirmation for sensitive requests. If a request arrives by email, verify it by phone or through an internal system. If it arrives by voice, confirm it through a separate channel. If it comes through a video call and involves money, access or confidential information, slow the process down. Schools, universities and workplaces need to teach verification as an everyday digital habit, not as a niche cybersecurity skill.

This is not about panic. but about recognising that the trust infrastructure of everyday life is changing. Phones, emails, video calls, dating apps, job platforms and social media were built around the assumption that people are difficult to imitate convincingly. AI is weakening that assumption.

The challenge is not to stop trusting everyone – that would make ordinary life impossible. The challenge is to stop treating familiar signals as proof. A voice can be copied. A face can be generated. A message can be polished. A profile can be manufactured. The familiar signals we once relied on are no longer enough on their own. In the age of scalable deception, the safest people will not be those who trust nobody. They will be those who know when to pause.