The Rise of AI-Powered Crypto Fraud: A Call for Action

Opinion by: Danor Cohen, co-founder and chief technology officer of Kerberus

Key Takeaways

In 2025, crypto fraud revenues reached an estimated $9.9 billion, with personal-wallet hacks accounting for 23% of thefts.
The use of AI in scams has evolved beyond novelty to practical applications for deception, including deepfake videos and voice replication.
The current security measures, such as audits and user alerts, are insufficient against the pace of AI-driven attacks.
A proactive approach is needed, integrating advanced defense mechanisms into transaction processes and smart contracts.
If the crypto industry fails to adapt and improve its security frameworks, it risks losing user trust and inviting regulatory intervention.

The Rise of AI-Powered Crypto Fraud

In 2025, the cryptocurrency landscape faces immense risks as artificial intelligence (AI) enhances the ability of scammers. Innovative techniques, such as deepfake technology and synthetic identities, are becoming mainstream methods used for deception. Despite the increasing sophistication of these scams, the industry's response remains inadequate. A shift from reactive measures to integrated security solutions is crucial to preserve trust within the crypto space.

A Changing Landscape: The New Era of AI-Driven Scams

As we step into 2025, the cryptocurrency sector finds itself in turbulent waters due to the rising tide of AI-enhanced scams. Techniques like deepfakes, voice synthesis, and simulated customer support agents have transitioned from experimental to essential tools for fraud. Last year saw crypto scams reach unprecedented levels, accumulating revenues close to $9.9 billion, with a staggering $2.17 billion already stolen in just the first half of 2025. Notably, compromises of personal wallets now represent nearly 23% of these thefts.

Despite this alarming rise in fraud, the industry's response remains limited to a familiar toolkit: audits, blacklists, reimbursement guarantees, user education campaigns, and retrospective reports. These responses are largely reactive, slow, and poorly equipped to address a threat that evolves at speeds more akin to a machine than to humans.

AI: The New Alarm for the Crypto Industry

The infiltration of AI into scams marks a pivotal shift in the methods used for deception. From simply enhancing the scale of scams, AI elevates the speed and tailored nature of attacks. Scammers can now swiftly replicate trusted settings or impersonate known individuals, making immediate defensive actions essential. Regulatory bodies, such as the Monetary Authority of Singapore, have begun to acknowledge these risks, issuing advisories regarding deepfake threats within the financial sector.

However, while the threat landscape has evolved, the industry's security mindset has stagnated, relying on outdated protective measures.

Outdated Security Practices Leave Users Vulnerable

Traditional security protocols in cryptocurrency prioritize static defenses like audits, bug bounty programs, and code reviews. While useful for identifying software vulnerabilities, these methods fall short against sophisticated social engineering that exploits human behavior. AI tools are increasingly employed not just to exploit behavioral weaknesses, but also to probe for and exploit coding flaws in a wide swath of contracts.

This dual vulnerability—technical flaws and human error—creates a precarious situation. Once attackers gain knowledge of security measures like blocklists, they can easily create new wallets or phantom domains for their schemes. With every audit and review, users fall victim to active exploits, and when incidents are labeled as "user error," it often absolves the system designers of accountability for systemic weaknesses.

In the traditional finance space, banks can freeze or reverse suspicious transactions, providing a safety net for consumers. However, in the crypto realm, once a transaction is signed, it's considered final, which heightens the risk when fraud occurs rapidly. Many security advisories, such as "Don’t click unknown links" or "Verify addresses carefully," become ineffective when the attacks originate from sources that users deem trustworthy.

Designing Systems for Proactive Protection

It’s time for the crypto industry to transition from reactive security measures to proactive design. The goal should be to develop systems capable of detecting anomalies in real time and intervening before the damage occurs. For instance, wallets could implement additional confirmations, temporarily hold transactions, or analyze user intent to mitigate risks effectively.

Infrastructure should encourage the sharing of intelligence among wallets, nodes, and security providers to create a network of threat signals, reputation scores, and anomaly indicators. This collaborative approach would prevent attackers from seamlessly slipping through security gaps.

In addition, incorporating contract-level fraud detection frameworks could identify potentially harmful behaviors in smart contracts. Shifting these protective measures from a retrospective stance to proactive user workflows is critical. It’s not about deploying AI universally but rather automating response mechanisms and ensuring a consensus on risk at every stage of a transaction.

The Need for Systematic Change

The crypto sector must take the initiative to implement systemic protections or risk being subjected to stringent regulations that could stifle innovation. Regulators are already moving towards overseeing financial integrity as part of algorithmic governance, which could result in restrictive frameworks if the industry does not preemptively adapt.

Moving From Defense to Assurance

It is our responsibility to restore confidence in the crypto space. The aim should not be to eliminate hacks altogether, but to significantly reduce irreversible losses. We need behavior that resembles "insurance-level" transactions—ones that are actively monitored and include checks, anomaly detection, and threat intelligence as built-in features.

The dialogue around security must evolve. While self-custody is essential, it alone cannot guarantee safety. It’s crucial to recognize that security tools should not be viewed as optional; they must be standard. Although education is important, intelligent design is paramount for safeguarding assets.

Our focus should not simply be on transaction speed or returns, but on building fraud resilience. Innovation should prioritize how effectively we can prevent harmful transactions rather than just the speed of blockchain settlements.

AI has illuminated the vulnerabilities in crypto’s security systems. The more significant challenge lies not in outsmarting these scams but in our unwillingness to evolve as an industry. The solution is not to hardwire AI into every application but to create systems that render AI-enabled fraud unfeasible for scammers.

If the crypto community remains reactive—merely issuing postmortems and pointing fingers at users—the cycle of deception will persist. The industry doesn't need to outwit AI in every engagement; it must mature and embed trust deeply within its operational fabric.