Are you being 'botfished'? A cybersecurity expert's advice for how to spot an AI dating scam

Does your cat like lasagna?

It’s a question one cybersecurity expert recommends people ask if they think they’re speaking with a bot on a dating app or website.

Robert Falzdon of Software Technologies Inc. Canada joined CTV Morning Live on Tuesday to warn people about increasing rates of so-called “botfishing.”

He says in the last year alone, Canada had a 20 per cent increase in scams involving bots, often being powered by artificial intelligence, which try to coax personal and financial information out of people looking for love online.

One explanation for the worrying trend is the massive online shift that happened as a result of the COVID-19 pandemic, when more Canadians were looking for virtual ways to connect and socialize with others.

“Unfortunately, scammers have noticed that, and as they’re noticing that, they’re starting to use some of these advanced tools,” Falzdad said.

Some of these tools include voice simulators, face generators, deepfakes—in which believable but fake photos or video are created—and chat bots that can create human-like text responses.

“(Bots) are using this really great, eloquent language that can be generated really quickly to get people to fall into that trap and to trust trust them and then start providing information,” Falzdon said.

As AI technology advances, it’s getting more and more difficult to spot the difference between a potential romantic partner and a bot.

“The ability of generative AI to create very, very realistic images of people is at the point where they’re almost indistinguishable by most experts,” said Falzdon.

Conducting a reverse image search of photos isn’t likely to produce results, as Falzdon says they’ve likely been created for the specific purpose of a dating profile.

However, Falzdon says there are a few steps people can take if they think they’re being scammed.

“First of all, look for language that looks a little too perfect. Sometimes the language can be a little too eloquent, especially when you’re speaking in an informal way to someone you just met,” he explained.

In addition, if the supposed person on the other side of the screen sends quick and lengthy responses to messages, that could also be a sign you’re speaking with a bot.

“Often they’ll lack specific personal details and if the responses don’t have those things in there, that’s often a sign that AI penned that,” Falzdon said, adding bots tend to overuse general knowledge, unprompted.

He suggests trying to throw the bot off by bringing up random topics out of context, including inquiries about a pet’s pasta preferences.

“Often the bot will have a difficult time or respond with something that doesn’t make sense either,” said Falzdon, although he failed to address the fact that a real human may be thrown off by out-of-context questions too.