The days of spotting a scam by its terrible spelling are over. Security experts are warning that criminals are now using AI “agents” to do the work that once needed a team of skilled fraudsters. These are software tools that can research, write and hold conversations largely by themselves.
In a report published by The National on Saturday, specialists from FTI Consulting, Kaspersky and the cybersecurity firm ParamountAssure set out how the technology is making cyber crime cheaper, faster and easier to scale. They also explained what ordinary people can do about it.
From chatbot to accomplice
Until recently, most worries about AI and crime focused on chatbots helping scammers write a convincing email. The experts say that has moved on. AI agents can now research a target through social media, professional profiles and company websites. They can tailor messages to that person, remember what was said earlier in a conversation, and juggle several victims at once.
“Agents are now doing the work, not just advising,” Farid Zahran, a senior managing director at FTI Consulting, told the newspaper. Premchand Kurup, chief executive of ParamountAssure, said tasks that once required expertise, time and a team can now be done with far fewer resources.
There are limits. Humans still choose the targets, oversee the attacks and control how stolen money is collected, the experts said. AI does not magically give criminals access to your private messages or bank records. Vladislav Tushkanov of Kaspersky’s AI Technology Research Centre cautioned that producing convincing content is not the same as reliably impersonating a real person without a human pulling the strings.
Real cases already on record
The warnings are not just theoretical. The article points to cases documented by AI company Anthropic. In one, an attacker used its Claude Code tool in an extortion campaign against at least 17 organisations, including healthcare providers, emergency services and government bodies, before Anthropic disrupted the activity. Some ransom demands topped $500,000, according to Zahran. In another case, a criminal with only basic coding skills used AI to build ransomware that was then sold to other criminals for between $400 and $1,200. A further Anthropic report in September described more cases where AI assisted with attacks and data theft, with human operators selecting targets and reviewing stolen data.
Microsoft’s latest Digital Defence Report, also cited, ranks the United States as the most targeted country for cyber attacks, followed by Israel, Ukraine, Taiwan, the UK and the UAE. Ireland does not appear in the top 12. But our nearest neighbour sits fifth, and Irish consumers have faced a steady stream of phone, text and parcel scams this year.
The new warning signs
Because AI writes fluent, polished text, the old giveaways no longer work. The experts suggest watching for behaviour instead:
- An unexpected request for money, especially from someone you know
- Pressure to act right now, before you have time to think
- A message saying bank or payment details have “changed”
- Any request for a password or a one-time verification code
Kurup warned that a familiar voice or even a video call is no longer proof that you are talking to the person you think you are.
How to protect yourself
The advice is simple but effective. If a request seems unusual, check it through a channel you already trust: a phone number you have saved, the official app, or someone you know at the organisation. Never use the contact details in the suspicious message itself, Tushkanov said.
If you have already sent money or clicked a link, contact your bank or payment provider immediately, then report it to the Gardaí. Keep the messages, screenshots and payment references. Change any passwords you may have shared or reused, and get technical help if you think a device has been compromised. One more tip: do not keep chatting with the scammer in the hope of getting your money back.
What happens next
The experts quoted by The National describe a trend that is speeding up as AI tools become cheaper and more capable. AI firms such as Anthropic have published reports on misuse and say they have disrupted some of the campaigns they found. But much of the burden still falls on ordinary users. For now, the best defence is old-fashioned: slow down, verify independently, and treat any urgent demand for money or codes as suspicious until proven otherwise.
Sources
- The National: AI agents are helping criminals to commit cyber crimes more cheaply and efficiently, say experts (31 October 2026)
Featured image: photo by Hamed Taha on Unsplash.
Leave a Reply