AI vs. AI: How Emerging Technologies are Battling Advanced Crypto Scams

The cryptocurrency sector is facing an alarming surge in scams, with U.S. citizens losing $9.3 billion to crypto scams last year, according to the FBI. This problem has worsened with the introduction of artificial intelligence (AI), which has led to a staggering 456% rise in AI-enabled scams in 2024. As bad actors adopt advanced technologies like generative AI, the cryptocurrency industry is implementing machine learning defenses to detect and prevent fraud. This article explores how both sides of this battle are evolving.

Key Takeaways

Scam Surge: The FBI reports that U.S. citizens suffered $9.3 billion in losses due to crypto scams last year.
AI’s Role in Fraud: AI technology has resulted in a significant increase in sophisticated scams.
Adaptive Scams: Generative AI is enabling criminals to execute scams on an unprecedented scale.
AI Defenses: The crypto industry is leveraging AI and machine learning to combat these frauds proactively.
User Awareness: Individuals can take measures to protect themselves from scams, such as being vigilant about website authenticity.

AI Scam Graphic

Increasing Sophistication of Scams

Scams are at the forefront of illegal activities in the cryptocurrency world. The FBI's findings indicate an astounding $9.3 billion was lost by U.S. citizens to these scams last year, with the problem exacerbated by the rise of artificial intelligence. TRM Labs, a blockchain analytics firm, notes that AI-facilitated scams increased by 456% in 2024 compared to prior years.

Deepfake Analysis

Advancements in generative AI (GenAI) have allowed bad actors to employ highly sophisticated technologies such as chatbots, deepfake content, cloned voices, and automated scam token networks at an unprecedented scale. Consequently, crypto fraud is shifting from being primarily human-driven to algorithmic operations that are swift, adaptable, and increasingly convincing.

Scams at High Speed

Ari Redbord, the global head of policy and government affairs at TRM Labs, shared insights with Cryptonews, stating that generative AI models can simultaneously launch thousands of scams. "We are witnessing a criminal ecosystem that is smarter, faster, and infinitely scalable," he remarked.

He elaborated on how GenAI can analyze a victim's language, location, and digital footprint. For example, during ransomware attacks, AI is used to identify the victims most likely to pay, draft ransom demands, and handle negotiations automatically.

In terms of social engineering, Redbord pointed out that deepfake technology is employed for scams like "executive impersonation" and "family emergency" situations, wherein deepfake voices and videos deceive both individuals and organizations.

Additionally, scams involving AI can transfer funds across numerous wallets within seconds, effectively laundering money at speeds unimaginable for human operations.

AI-Powered Protective Measures

In response to the rise in AI-facilitated scams, the cryptocurrency industry is developing AI-driven defenses. Blockchain analytics firms, cybersecurity companies, exchanges, and academic researchers are collaboratively creating machine learning systems aimed at detecting, flagging, and mitigating fraud before victims lose money.

For instance, Redbord mentioned that AI is integrated into every layer of TRM Labs' blockchain intelligence platform, utilizing machine learning to analyze trillions of data points across over 40 blockchain networks. This enables TRM Labs to visualize wallet connections, identify fraud patterns, and highlight suspicious behaviors indicative of illegal activities.

“These systems not only recognize patterns but also learn from them. As data evolves, so do the models, adapting to the constantly changing crypto landscape,” Redbord explained.

This capability allows TRM Labs to uncover thousands of minor, seemingly unrelated transactions that may signify a scam, laundering scheme, or ransomware operation.

Sardine, an AI risk platform, is taking a similar route. Founded in 2020, during the early days of major crypto scams, the company employs an AI-fraud detection process composed of three layers.

“Our data-centric approach captures deep signals from every user session on crypto exchanges, along with device attributes, tampering indicators, and user behavior,” shared Alex Kushnir, Sardine’s head of commercial development. The platform also integrates a broad network of reputable data providers and uses consortium data for sharing intel on bad actors among companies.

Kushnir noted that Sardine utilizes a real-time risk engine that acts on these signals to address scams as they emerge.

Effective AI Strategies

The tools employed in the ongoing battle against scams are proving effective. Matt Vega, Sardine’s chief of staff, mentioned that once a pattern is recognized, their AI conducts a comprehensive analysis to derive trend recommendations, significantly reducing the time compared to human analysis.

For instance, Sardine collaborates closely with major crypto exchanges to identify unusual user behavior. Transactions analyzed by Sardine’s decision platform use AI diagnostics to proactively notify exchanges of potential issues.

A recent blog post from TRM Labs revealed an instance where they encountered a live deepfake during a video call with a suspected financial grooming scam perpetrator. Scammers often work to build long-term relationships with victims to secure access to their funds.

“We believed this scammer was utilizing deepfake technology, evidenced by the unnatural appearance of their hairline,” Redbord explained. AI detection tools successfully confirmed that the image was likely AI-generated. Despite the successful detection, this scam and others like it have already defrauded victims of approximately $60 million.

Kidas Detection

Cybersecurity firm Kidas is also applying AI for scam detection and prevention. Founder and CEO Ron Kerbs disclosed that as AI-driven scams have surged, Kidas’ proprietary models can evaluate content, behavior, and audiovisual inconsistencies in real-time, enabling the identification of deepfakes and AI-generated phishing attempts.

"This facilitates instant risk scoring and real-time interdiction—essential for counteracting automated fraud operations," said Kerbs. Just recently, Kidas successfully intercepted two distinct crypto scam attempts on Discord, showcasing its effective real-time behavioral analytics.

Ongoing Battle Against AI-Enabled Scams

While AI-based tools have shown promise in detecting and preventing advanced scams, the frequency of these attacks is expected to persist.

"AI is lowering the entry barriers for sophisticated crimes, creating scams that are both scalable and highly personalized, which will attract more perpetrators," Kerbs warned.

He anticipates that semi-autonomous malicious AI agents will soon be capable of orchestrating entire campaigns with minimal human involvement, including untraceable voice-to-voice deepfake impersonations during live conversations.

Despite the potential risks, Vega emphasized that users can adopt specific measures to shield themselves from falling prey to these scams.

For example, he pointed out that many attacks occur through spoofed websites where users may click on fraudulent links.

"Look for unusual Greek alphabet characters in URLs. Recently, Apple fell victim to such a scheme where a scammer created a fraudulent site using a Greek ‘A’ in their name. Additionally, users should avoid sponsored links and scrutinize URLs closely."

Furthermore, companies like Sardine and TRM Labs are actively collaborating with regulators to develop strategies that leverage AI to reduce the risk of AI-driven scams.

“We are creating systems that provide law enforcement and compliance officials with the speed, scale, and reach that criminals currently possess—from real-time anomaly detection to identifying coordinated cross-chain laundering. AI is enabling a shift in risk management from reactive to predictive,” Redbord concluded.