AI has arrived, and it’s already changing things in the world of crypto. Coders use it to code, researchers use it to research and, unfortunately, scammers use it to scam. That is the finding of a new report by blockchain analytics firm Elliptic about the emerging risks of AI in perpetuating criminal use of crypto.
This is an excerpt from The Node newsletter, a daily roundup of the most pivotal crypto news on CoinDesk and beyond. You can subscribe to get the full newsletter here.
Note: The views expressed in this column are those of the author and do not necessarily reflect those of CoinDesk, Inc. or its owners and affiliates.
“The rise of artificial intelligence has shown huge potential for driving innovation, not least within crypto. However, as with any emerging technology, there remains a risk of threat actors seeking to exploit new developments for illicit purposes,” the report reads.
While the risk right now remains small, the firm’s researchers did identify five “typologies” where AI is already being deployed in nefarious ways. These include in creating and disseminating deepfakes to make more convincing scams, building AI-scam tokens to capitalize on hype, using large language models to devise hacks, spreading disinformation and making more convincing phishing websites/prompts to facilitate identity theft.
Awareness of these new (or frankly, old, but now supercharged) scams means that users can stay ahead of the curve. That means crypto users should become more familiar with the most common types of crypto-related scams. CoinDesk has a good report on that front here covering all the basics like social media scams, Ponzi schemes, rug pulls and “romance scams” (now often referred to as “pig butchering”).
“The reason there is no easy way to deal with the problem is because it’s really multiple problems, each with its own variables and solutions,” Pete Pachal, author of the excellent Media CoPilot Substack, wrote in a recent piece about deepfakes, AI and crypto.
According to Pachal, who recently spoke at a Consensus 2024 session called “From Taylor Swift to the 2024 Election: Deepfakes vs. Truth,” deepfakes have become increasingly difficult to spot as AI image generation has improved. For instance, earlier this month a video circulated on social media of fake Elon Musk promoting fake trading platform Quantum AI that promised users fake returns that apparently tricked more than a few people.
Instances like these are likely only going to grow. Verification company Sumsub claims that crypto was “the main target sector” for almost 90% of deepfake scams detected in 2023. While it’s unclear how effective these scams were, the FBI’s online crime report found crypto investment losses in the U.S. grew 53% to $3.9 billion last year.
However, it’s worth noting that oftentimes instances of fraud in the crypto industry are just incidentally related to crypto, because it just happens to be a topic that draws a lot of attention and is often complicated for people not steeped in the culture.
As CFTC Commissioner Summer Mersinger told CoinDesk: “I think it’s a little unfair because a lot of these cases are just run of the mill fraud; somebody stealing someone else’s money, someone claiming to buy crypto, but not actually buying the crypto. So we’ve seen this play out with whatever the hot topic is at the time.”
If there’s any consolation, it’s that images, video and text generated by AI are still relatively easy to notice if you know what to look for. Crypto users in particular should be vigilant, considering how common it is for even high-profile figures to get tricked by social engineering schemes or malicious scripts.
MetaMask builder Taylor Monahan has sage advice here: always know you’re a potential target, and actually verify what you’re clicking on is what it purports to be.
Crypto is already a low-trust environment, just given the nature of the technology. And it might get even lower.
