In brief
- AI deepfakes of celebrities, government officials, and more accounted for 40% of “high-value” crypto fraud in 2024, a Bitget report claimed.
- Crypto scam losses reached $4.6 billion in 2024, representing a 24% increase from previous year.
- Social engineering schemes ranked as second-most dangerous crypto fraud method, followed by modern Ponzi schemes.
AI-generated deepfake impersonations of government officials, billionaires, and celebrities accounted for 40% of “high-value frauds” in 2024, a Bitget report revealed.
That same year, $4.6 billion in crypto was lost to scams, a 24% increase from the previous year, Bitget’s Anti-Scam Report 2025—co-authored with Slowmist and Elliptic—found.
“Crypto scams have entered a new era—driven by AI deepfakes, social engineering, and deceptive project fronts,” the report said. “Scams now exploit trust and psychology as much as technology. From wallet takeovers to multimillion-dollar frauds, the attacks are becoming more personalized, more believable, and harder to detect.”
A “frequent” deepfake that appeared, according to the report, featured Tesla CEO Elon Musk pitching fraudulent investment or giveaway schemes. Other uses for deepfakes include bypassing know-your-customer verification, creating virtual identities to lead investment scams, and conducting fake Zoom phishing attacks.
The Zoom scam sees scammers impersonate executives, experts, and journalists—including Decrypt reporters—to get a victim on a fraudulent video call. This may see the attacker offer the victim a job or ask to interview them for a non-existent article. During the fake Zoom call, attackers can gain control of the victim’s computer, steal data, and potentially access private crypto keys.
Bitget reports that in some cases, the attackers are using deepfake tools to generate video and audio content to trick the victim into joining the call.
Many of these deepfake scams aren’t new; the Elon Musk scam first went viral in 2022. However, with the rapid acceleration of artificial intelligence, deepfakes have started to look real. So much so that President Trump signed the bipartisan Take It Down Act last month that protected the victims of deepfake porn—although, deepfakes in general still aren’t prohibited.
In May, actor Jamie Lee Curtis called out Meta CEO Mark Zuckerberg after discovering a fake AI-generated ad of herself promoting a product without her consent.
“The biggest threat to crypto today isn’t volatility—it’s deception,” Bitget CEO Gracy Chen said, in a release. “AI has made scams faster, cheaper, and harder to detect.”
Outside of deepfakes, social engineering and modern Ponzi schemes ranked second and third as the report’s “most dangerous” scams.
By definition, social engineering scams exploit the psychology of victims in a method that is “low-tech, yet highly effective,” the report says. One of the most common examples is the pig butcher scam, also called a romance scam, which sees the attacker develop a relationship with the victim solely to scam them.
As for the ol’ classic Ponzi scheme—named after the early 20th century scammer Charles Ponzi—the report claims that the scam has gone through a “digital evolution.”
“These scams typically disguise themselves in new concepts like DeFi, NFTs, and GameFi, packaged as project fundraising, liquidity mining, or platform token staking,” the report states. “Fundamentally, they remain classic Ponzi schemes where ‘new money fills old holes.’ Once the cash flow breaks or the operators cash out and exit, the entire system collapses quickly.”
The report notes a rise in Ponzi schemes that adopt “game-like user interfaces” and use deepfakes to forge endorsements from celebrities to boost credibility in the scheme. AI has revolutionized the scamming industry, resulting in a markedly different landscape from just a few years ago.
“Five years ago, avoiding scams meant ‘don’t click suspicious links,'” the report finished. “Today, it’s ‘don’t trust your own eyes.'”
Edited by James Rubin
Daily Debrief Newsletter
Start every day with the top news stories right now, plus original features, a podcast, videos and more.