Amidst crypto scams surging 456% in 2024, AI-driven defenses from TRM Labs, Sardine, and Kidas are using machine learning to outsmart next-gen cybercriminals.
As artificial intelligence reshapes industries, it is also transforming one of the world’s most profitable criminal enterprises: cryptocurrency scams. But now, a new technological arms race is underway — where AI is being deployed not only to commit fraud but also to stop it.
According to the U.S. Federal Bureau of Investigation (FBI), American citizens lost a staggering $9.3 billion to crypto scams in 2024. The surge in AI-generated fraud, driven by deepfakes, cloned voices, and synthetic social media accounts, has made digital deception more convincing and more difficult to detect.
Blockchain analytics firm TRM Labs reported a 456% increase in AI-facilitated scams last year, as generative AI tools made it possible for criminals to automate phishing, impersonation, and money-laundering operations at unprecedented speed.
“We’re seeing a criminal ecosystem that is smarter, faster, and infinitely scalable,” said Ari Redbord, TRM Labs’ global head of policy and government affairs.
AI-powered fraudsters now use advanced models to tailor scams to victims’ language, habits, and online activity. In some ransomware cases, AI helps select victims most likely to pay — even automating ransom negotiations. Deepfake voices and videos, meanwhile, have been used in “executive impersonation” and “family emergency” scams that target both individuals and companies.
Read Also: OpenAI Strikes $38B Cloud Deal With Amazon To Boost AI Power
To combat this evolving threat, major crypto and cybersecurity firms are developing their own AI defenses. TRM Labs has embedded machine learning into its blockchain intelligence network, scanning trillions of data points across more than 40 blockchains to flag suspicious patterns invisible to human analysts.
“These systems don’t just detect patterns—they learn them,” Redbord explained. “As the data evolves, the models evolve too.”
Fraud-detection firm Sardine uses a three-layer AI system that analyzes device behavior, taps external data providers, and shares insights through an anti-fraud consortium. Its models act in real time, blocking fraudulent transactions as they occur.
“Machine learning remains the gold standard in predicting risk,” said Alex Kushnir, Sardine’s head of commercial development.
Cybersecurity startup Kidas is also employing proprietary AI models to spot real-time deepfake and phishing attempts. Its founder, Ron Kerbs, said Kidas recently intercepted two large-scale crypto scams on Discord by analyzing behavioral anomalies and audio-visual inconsistencies.
Kerbs warned that “semi-autonomous malicious AI agents” could soon run entire fraud campaigns with little human control, creating a global challenge for regulators and security professionals alike.
Still, experts believe that AI-powered defense systems — built by firms such as TRM Labs, Sardine, and Kidas — can turn the tide. These platforms are working with regulators to design predictive models that give law enforcement the same scale and speed as cybercriminals.
“We’re building systems that make risk management proactive, not reactive,” Redbord said.
The battle between artificial intelligences — one malicious, one defensive — may define the future of digital finance. For now, it is AI versus AI in the world’s newest frontier of cybercrime.