From dipfakes to fake voices: how not to fall for AI scams

When Gabriel Aguilar was a student, he himself fell victim to fraudsters pretending to be employers. After a "job interview", they sent him a fake cheque for the purchase of a laptop, asking him to return the change. Fortunately, the bank suspected the forgery in time. But now, years later, Aguilar - now an assistant professor of technical writing at the University of Texas at Arlington - is studying how such schemes could become far more dangerous in the age of artificial intelligence.

AI for evil: new tools - old tricks

In his new paper published in the Journal of Business and Technical Communication, Aguilar analyses how today's fraudsters are using AI - from voice bots to dipfakes - to make traditional deception schemes even more convincing.

"AI hasn't reinvented fraud," the researcher notes. - It has simply made the old tricks much more realistic."

He places particular emphasis on vulnerable populations - particularly the US Hispanic community, which scammers often target. Their gullibility, language barriers and lack of digital literacy make such communities particularly vulnerable to AI-based deception.

How to protect yourself from digital threats

Aguilar suggests using technical writing courses as a tool to build AI literacy and recognise fraud. Such courses teach students to present complex information clearly and accurately, follow instructions and make informed decisions - skills vital in a world where plausible lies can be generated in seconds.

Technical writing (or technical writing in English) is a type of professional written communication that aims to convey specialised information clearly, understandably and accurately. It is used in science, technology, business, IT and other fields where complex things need to be explained simply.

What technical writing includes:

• Instructions and user manuals (such as how to assemble furniture or customise software)

• Technical reports and documentation

• Process and proceduredescriptions (e.g., for manufacturing or science labs)

• API documentation and programming references

• Business communication: business letters, proposals, grant applications, project summaries

• Popular science texts for general audiences based on complex data

Why is it necessary?

• To convey complex information clearly: from an engineer to a manager, from a scientist to a layman, from a developer to a user.

• To prevent errors and misunderstandings in technical processes.

• To ensure safety and efficiency, especially in medicine, manufacturing, aviation and IT.

• To increase digital literacy, as in the case of a University of Texas professor who uses technical writing courses to teach students to recognise AI-based fraudulent schemes.

An example of a simple application:

Bad:

Connect the device, following ISO 9001 regulations taking into account the voltage parameters described in the CE certificate application.

Good (technical letter):

Plug the power cord into the power socket. Make sure that the voltage corresponds to 220 V. Do not use the device without grounding.

In his teaching methodology, he has developed a four-step model to enable students to understand how AI can be used in fraud and to recognise danger signals in time. The key, according to the researcher, is to foster critical thinking and teach students to share knowledge with their families and communities, increasing "AI literacy" from the bottom up.

"When people start noticing small inconsistencies," Aguilar says, "they are no longer so easily manipulated. It's an awareness-based defence."