AI-Powered Crypto Scams: Understanding the Threat and How to Safeguard Yourself

AI-Powered Crypto Scams: Understanding the Threat and How to Safeguard Yourself

The cryptocurrency industry is increasingly vulnerable to scams, particularly as artificial intelligence (AI) technologies become more sophisticated. These scams are not only more believable but also more widespread, utilizing advanced methods like deepfakes and automated bots. Awareness and education are crucial in combating these threats.

Key Takeaways

There has been a significant rise in AI-driven crypto scams, with reports increasing by 456% year-on-year.
AI scams often target financially active cryptocurrency users, using advanced techniques like machine learning for deception.
Prompt injection attacks present new vulnerabilities, allowing attackers to manipulate AI systems for malicious purposes.
It is essential for users to stay vigilant and verify sources when engaging with crypto content or opportunities.

The Rise of AI-Powered Crypto Scams

The cryptocurrency field is notorious for hacks, and the recent growth of artificial intelligence (AI) has magnified the risk of fraud and scams. What were once basic scams have evolved into highly convincing and scalable threats due to advancements in generative AI (genAI), deepfakes, voice cloning, and automated bots.

Alarming Growth in AI Scams

This year has seen a striking increase in AI-driven scams. Data from Chainabuse, an open-source fraud reporting platform by TRM Labs, shows that from May 2024 to April 2025, reports of genAI-enabled scams surged by 456%. This followed a previous increase of 78% during the 2023-2024 period.

Data from TRM Labs on AI scams

According to statistics from Chainalysis, about 60% of all deposits into scam wallets can be traced back to AI-driven scams, a figure that has risen steadily since 2021.

Chainalysis statistics on scams

What Exactly Are AI-Powered Crypto Scams?

Eric Jardine, cyber crimes research manager at Chainalysis, indicates that AI-powered scams are transforming the landscape of crypto crime. While these scams can target anyone, they typically attract those actively participating in cryptocurrencies but who are less familiar with the schemes involving AI.

“Criminals are leveraging the anonymity of digital assets combined with AI to exploit users on a larger scale,” Jardine stated. “These scams can create fake identities, realistic conversations, and websites or apps that closely resemble legitimate platforms.”

A recent example involved scammers creating a fake YouTube livestream that mimicked NVIDIA’s annual GTC event, showcasing an AI deepfake of CEO Jensen Huang. They falsely claimed that NVIDIA was endorsing a crypto project to entice viewers into investing.

Common types of AI-powered scams include:

Deepfake videos of credible public figures promoting dishonest crypto projects.
Fraudulent automated trading platforms that promise unrealistic returns.
AI-generated phishing websites designed to steal sensitive information.
Voice cloning to impersonate executives or loved ones in order to deceive victims.

AI-generated advertisements have also been exploited to fool cryptocurrency investors. For instance, in Spain, a group executed a scam using AI-created ads featuring fake celebrity endorsements, leading to the arrest of six individuals who defrauded over 200 victims out of approximately 19 million EUR.

Nick Smart from Crystal Intelligence warns that the frequency of these scams is startling and rapidly increasing. “AI has made it easier for anyone to execute sophisticated scams; you no longer need to be a tech expert to pull off a convincing operation,” he said.

Smart also pointed out the profitability of such scams, referencing a previous deepfake scam featuring Elon Musk that generated $10,000 from one scheme, while link analyses indicated that the scammers earned as much as $1 million from multiple operations they shared.

The Rise of Prompt Injection Attacks

The crypto environment is also facing an increase in prompt injection attacks. This security vulnerability allows attackers to manipulate AI agents or large language models (LLMs) to carry out harmful actions.

“Prompt injection attacks are a new method where an attacker inputs seemingly legitimate content designed to provoke unintended behaviors in machine learning models,” Smart explained. “It’s like tricking your ChatGPT into doing something it shouldn’t.”

As LLMs connect with other services—like crypto wallets, web browsers, and social media—these prompt injections could become even more problematic. For example, skilled attackers could use them to implant fake memories into an AI model, which could then transfer sensitive discussion histories to an attacker-controlled email.

A blog post from Brave Browser highlights the potential dangers of AI-powered browsers. If users are logged into sensitive accounts while using these tools, even innocuous tasks like summarizing a post could lead to significant data breaches.

In the context of cryptocurrency, harmful AI agents might gain control over users' wallets and make unauthorized transactions. For instance, if an attacker successfully inputs a prompt instructing the AI to divert funds to their wallet, victims could incur substantial losses.

Protecting Yourself from AI-Powered Scams

Experts caution that AI-powered scams are likely to keep evolving. “As AI technologies grow more sophisticated and accessible, scams are becoming more realistic, quicker to implement, and increasingly difficult to spot,” Jardine noted.

Smart emphasized that newcomers to crypto should be particularly cautious, as they may lack familiarity with common scam tricks. Wealthy individuals and institutions are also prime targets for tailored sophisticated attacks.

However, users can adopt several strategies to enhance their security:

Verification is Key: If you see a celebrity endorsing a crypto opportunity, approach it with skepticism until you can verify its legitimacy. Always navigate to official websites instead of clicking on provided links.
Utilize Detection Tools: Tools such as Crystal Intelligence’s scam-alert.io allow users to check wallet addresses before sending any funds. Furthermore, Chainalysis employs AI to improve entity recognition and analyze blockchain activities for signs of fraud.

Jardine mentioned Chainalysis's AI-powered fraud detection solution, Alterya, which identifies scammers before they can reach potential victims. Alterya employs machine learning models and deterministic data to assess recipient address risks and help minimize fraudulent transactions.

Ultimately, increased awareness and vigilance are the best defenses. “While AI tools like Alterya aim to prevent scams from proliferating, users must still verify sources and remain alert to unsolicited communications,” Jardine concluded.