What do Taylor Swift and Joe Biden have in common? Not much, it seems, but in recent weeks they have both been the targets of the malicious use of deepfake technology. The technology, which uses artificial intelligence (AI) to mimic a person can make it seem audio or video footage is them, saying or doing things they wouldn’t ever do.
In late January, the social media platform X – formerly known as Twitter – had to halt searches for Taylor Swift after AI-generated explicit images slipped through the platforms content filters and went viral. Unable to stop its spread, the platform implemented a ban on the search term ‘Taylor Swift’, an unsatisfactory and temporary fix to what is a growing problem. “If it can happen to Taylor Swift, it can happen to you,” Carolina Are, a fellow at Northumbria’s University Centre for Digital Citizens in the UK, was reported by TechCrunch as saying.
US President Joe Biden also became the target of the malicious use of this technology, highlighting how disinformation poses a major risk to the upcoming presidential elections in the US. A deepfake version of Biden’s voice had been used in robocalls – where automated phone calls are made using AI – discouraging Democrats from voting in the New Hampshire primary elections. Although it was a spoof of Biden’s voice, it sounded remarkably like him.
Given how so much of the working day is conducted over the phone and Zoom, the potential of this increasingly-sophisticated technology is scary. For now, video deepfakes are harder to create and so the risk corporate treasurers are facing is in audio deepfakes. And because they have access to the company purse strings, they are most likely to be targeted for fraud.
As has been previously reported in a feature for Treasury Today magazine, there have been a couple of cases that should sound the alarm bells for many treasurers. One corporate banker started to transfer US$35m on the request of a senior director of a corporate based in the UAE. This executive had supposedly called the company’s main banker in Hong Kong informing them they were about to purchase a company and needed to transfer the funds to complete the transaction. The banker concerned knew the person they were speaking to, and they had seen emails mentioning a lawyer, who the executive mentioned during the fateful phone call. The banker took the phone call at face value and started to transfer the funds. It later transpired the whole thing was a scam and was part of sophisticated money laundering network using AI to enable their crimes.
There was another case in 2019 where a UK-based energy company also fell for a voice on the other end of the phone and a managing director transferred US$243,000 on the instructions of the CEO, a German called Johannes. It was, however, not Johannes they were speaking to. The company’s insurance company reported, “The software was able to imitate the voice, and not only the voice: the tonality, the punctuation, the German accent”.
And these are the cases that have made it into the news; it can be assumed many other such attempts have been made and have been kept quiet.
Panda Security notes the main business threat of deepfake technology is fraud. There is also the risk of fabricated comments where a senior leader is saying or doing something they shouldn’t. Another risk is extortion, where an explicit image is created and paying up stops it from going viral. While for most people, the Taylor Swift and Joe Biden incidents may seem far removed from their work lives, the growing sophistication – and availability – of the technology means that someday it could happen to you.
