Good morning. I’m Alexandra Posadzki. My colleague Joe Castaldo and I dug into a story that might make you second-guess whether it’s wise to record your own voice-mail greeting.

Scammers are using deepfakes generated by artificial intelligence to defraud businesses, creating a new financial risk for companies to manage. In one high-profile instance, an employee was duped into sending fraudsters more than US$25-million. More on that below, but first:

Teck Resources Ltd. is open to a deal with Glencore around combining its copper operations in Chile.

The Nova Scotia government is moving to dismantle interprovincial trade barriers with new legislation.

The death of data: Under the new Donald Trump administration, key information is disappearing.

TFI International Inc. eyes move to the U.S. after the Canadian trucking company’s CEO said Trump’s administration will be great for business.

  • Canadian retail data will be released today, and Bank of Canada Governor Tiff Macklem will be speaking to the Mississauga Board of Trade-Oakville Chamber of Commerce about trade friction, structural change and monetary policy.

Photo illustration by The Globe and Mail/iStockPhoto / Getty Images

I started hearing about the rise in deepfake scams during meetings with cybersecurity executives shortly after taking on a new role focused on financial scams and cybercrime. Prior to this beat, I spent five years as The Globe and Mail’s telecom reporter.

In boardrooms, at conferences and over dinners, I kept hearing variations of the following story: A victim gets a call from a loved one who says they’re in trouble and need money – fast. But something seems off about the call. For one thing, it’s from an unknown number. The victim then contacts the friend or relative directly and learns that the call was fake.

One executive even played me the audio recording that had been left on his answering machine, purportedly by a friend of his calling from jail. He seemed unsettled by how convincingly AI had cloned his friend’s voice.

I teamed up with Joe Castaldo, who has written extensively about the promise of generative AI, to explore what’s behind the rise in deepfake scams.

Many people have seen or heard about the fake videos of celebrities such as Elon Musk and Justin Trudeau peddling fraudulent investment schemes. According to a report from Entrust Corp., a Minneapolis-based company that processes millions of identity verifications each year, the number of deepfakes increased by 3,000 per cent from 2022 to 2023.

What we learned through our reporting is that it’s not just consumers who need to be wary of deepfakes. Fraudsters are using generative AI to try to steal money from businesses as well, for instance by impersonating chief executives and other business leaders.

One of the most well-known examples occurred when scammers convinced a Hong Kong employee of British engineering giant Arup Group Ltd. to transfer them US$25-million of the company’s funds. The employee was initially skeptical, but fell for the ruse after attending a video call featuring deepfake recreations of several co-workers, including the company’s chief financial officer.

In the United States, the Financial Crimes Enforcement Network has warned financial institutions to look out for scammers who use deepfake technology to create fraudulent identification documents, such as drivers’ licenses and passports, and then use them to open accounts and launder money.

Deepfake scams that target businesses are still a relatively new phenomenon, but experts predict they’ll become more commonplace.

More than a quarter of the executives polled by Deloitte last spring said their organization had experienced at least one deepfake incident targeting financial data. And more than half of the respondents to a follow-up question said they expected the prevalence of such attacks to increase over the next 12 months.

The reason behind the growing popularity of these scams lies in how inexpensive and easy they are to create. You only need a short recording of someone’s voice in order to clone it, which is why some cybersecurity professionals say it can be dicey to record your own voice-mail greeting. Meanwhile, the tools to clone someone’s voice can be purchased very cheaply – or even accessed for free on GitHub.

The technology is improving as well, which means the clones are becoming more and more convincing.

Thankfully, it’s not all bad news. New startups are popping up to combat nefarious uses of generative AI, including deepfakes, and some of the startups seem promising. For example, software from DeepTrust, a San Francisco-based startup, can analyze video-calling platforms to detect AI-generated audio.

Experts told us that while technology created the problem, it’s also the key to solving it.

Unfortunately, when it comes to cybersecurity, the old cliché is often true: The bad guys tend to be one step ahead.

More from The Globe

  • AI-powered scams: Protect yourself in the new era of hyper-personalized phishing.
  • Opinion: Why Canadian students need a crash course in AI literacy.
  • Fact check with us: The difference between disinformation and misinformation, and how to do your own analysis.