44% of Crypto Platforms Flag AI Activity in Customer Onboarding

Artificial intelligence is already appearing in cryptocurrency identity verification systems, but not always in the way fraud statistics suggest.
Summary:
- Regula found that 44% of surveyed crypto firms had confirmed AI-assisted or automated activity in identity checks, compared with 31% in other industries.
- Document fraud was cited more frequently than deepfake impersonation, suggesting that conventional identity threats remain prominent.
- The emergence of customer-authorized AI agents creates new questions about consent, account access and responsibility for automated actions.
A survey published October 8 by identity verification provider Regula found that 44% of cryptocurrency companies had confirmed AI-assisted or automated activity on the customer side of identity checks. Across five other industries, the figure was 31%.
The results reflect an emerging problem for crypto exchanges and other digital-asset businesses: software can now interact with onboarding systems on behalf of customers, while malicious actors can use similar capabilities to submit manipulated identities or automate fraudulent applications.
Regula’s findings, based on research conducted by Sapio Research in March 2026, also suggest that crypto companies are comparatively well prepared to reconstruct identity verification decisions after an incident. That capability is increasingly relevant as regulators, auditors and customers demand explanations for automated approvals and account restrictions.
The October 8 announcement provides a snapshot of how the industry is responding, although its findings require careful interpretation.
What the 44% AI Detection Figure Actually Measures
Regula commissioned Sapio Research to survey 850 professionals responsible for fraud prevention and financial crime controls across seven markets. Of those respondents, 102 worked in cryptocurrency businesses, while 748 represented banking, financial services, telecommunications, government, and gaming and gambling.
The survey asked whether organizations had encountered AI-assisted or automated actors interacting with their identity verification systems during the previous 12 months.
Among the cryptocurrency respondents, 45 confirmed such activity through investigations or post-incident reviews.
That is the basis of the 44% headline.
It is not a measurement of how many customers used AI, how many verification attempts were fraudulent or how frequently automated attacks succeeded.
An exchange that detects one automated onboarding attempt and another that identifies thousands could both contribute equally to the proportion of organizations reporting confirmed activity.
Detection capabilities also influence the result. Businesses with more extensive monitoring and incident investigations may be more likely to recognize automation.
Other responses showed considerably smaller differences between industries. Uncertain scripted activity was reported by 34% of crypto firms and 35% of their peers. Attempted activity that had been blocked or contained was reported by 20% and 21%, respectively.
Those findings limit the conclusions that can reasonably be drawn from the headline comparison.
The survey documents confirmed encounters with automation, not proof that crypto identity systems are inherently more vulnerable.
Stolen Documents Remain a More Immediate Concern Than Deepfakes
Much of the attention surrounding AI-related identity fraud focuses on synthetic faces and manipulated videos. Regula’s crypto respondents, however, ranked conventional document fraud as a greater concern.
Some 44% identified counterfeit, altered or stolen identity documents as a significant threat, compared with 37% citing deepfake or AI-generated impersonation.
The difference reflects two separate verification problems.
Document authentication examines whether an identity document is genuine, whether its security features are intact and whether its information has been manipulated.
Biometric verification attempts to establish whether the person presenting that document matches its holder. Liveness checks add another layer by assessing whether the biometric evidence comes from a live person rather than a recording or synthetic presentation.
An applicant using a genuine stolen passport could pass a document-authenticity check while failing to establish legitimate ownership. Conversely, a convincing synthetic facial image might match the photograph on a genuine document while concealing the absence of the rightful holder.
A system relying on only one of these controls leaves the other problem unresolved.
Crypto firms also need to establish whether automated software is acting with a customer’s authorization. A legitimate user delegating administrative tasks to an AI assistant presents a different risk from an attacker deploying scripts to open accounts using stolen identities.
Blocking every automated interaction could interfere with legitimate activity. Allowing automation without checking its authority could weaken account protections.
The operational challenge is identifying the person responsible for an interaction and establishing whether the evidence presented can be trusted.
Crypto Firms Report More Complete KYC Audit Trails
The survey’s clearest difference between cryptocurrency companies and other sectors concerns the ability to explain identity decisions after they have been made.
Of the 102 crypto respondents, 63 said their organizations could fully reconstruct a verification decision, tracing the systems involved, the evidence reviewed and the logic used to approve or reject an applicant.
That represents 62%, compared with 49% across the remaining industries.
This capability matters when an exchange investigates suspicious account activity or responds to a customer challenging an identity-related restriction.
READ MORE: OCC’s Crypto Trust Charter Framework Heads to Federal Court
An approval timestamp and a successful verification result offer limited information. Investigators may also need the original document checks, biometric outcomes, automated risk signals and any manual intervention that affected the decision.
The survey found that 87% of crypto respondents had been required to explain an identity-related decision to a regulator, court or external auditor, compared with 81% of respondents elsewhere.
Among organizations receiving such requests, 70% of crypto respondents reported being able to produce audit-grade evidence, against 68% in other industries.
These are self-reported capabilities, not independently audited compliance outcomes. A detailed record can establish how a decision was reached without proving that the underlying controls were adequate.
Crypto firmsOther sectors
Share of respondents
44%
31%
44%
34%
62%
49%
87%
81%
Regula / Sapio Research, March 2026. Crypto respondents: 102; other industries: 748. Percentages reflect self-reported organizational experience, not measured fraud incidence. Differences are expressed in percentage points.
The Financial Cost of Getting Identity Checks Wrong
Regula also found that 48% of cryptocurrency respondents associated incorrect identity verification outcomes with financial losses, compared with 38% in other sectors.
For an exchange, a false approval can allow a fraudulent applicant to open an account, move assets or gain access to services under another person’s identity. A false rejection can prevent a legitimate customer from trading or withdrawing funds.
The consequences become particularly difficult when identity checks are repeated after an account has already received deposits.
A customer may pass initial verification, transfer cryptocurrency to the platform and later encounter restrictions following a new risk alert. If the original decision cannot be explained, the exchange may struggle to determine whether the restriction reflects newly discovered evidence or an earlier verification failure.
There is no published monetary estimate of these losses in the survey, and the reported percentages do not measure the frequency of incorrect decisions.
The financial significance lies in the competing risks: weak verification can expose platforms to fraud, while excessively restrictive controls can interfere with legitimate customer access.
Regula is a commercial identity verification provider, and the research covers only 102 crypto respondents. Its findings are therefore better understood as reported industry practices and concerns than as an independent assessment of the effectiveness of cryptocurrency KYC systems.
AI Agents Will Test the Limits of Existing KYC Rules
Customer-authorized AI agents introduce a question that traditional identity verification was not designed to answer: what happens when a legitimate customer delegates part of an interaction to software?
A person may authorize an assistant to prepare account information, upload documents or manage administrative requests. That authorization does not necessarily extend to changing withdrawal credentials, accessing sensitive records or initiating financial transactions.
Crypto platforms will need to distinguish these permissions rather than treating every automated interaction as either harmless or fraudulent.
The findings also come amid broader regulatory scrutiny of cryptocurrency compliance. In its July 2026 targeted update, the Financial Action Task Force (FATF) examined persistent weaknesses in virtual-asset supervision and increasingly sophisticated fraud risks. Although the assessment does not specifically measure AI-assisted identity verification, it provides a wider regulatory context for the challenges identified by Regula.
The Financial Action Task Force’s guidance on digital identity provides a relevant regulatory foundation. It emphasizes reliable, independent identity evidence and a risk-based approach to customer due diligence, including for virtual asset service providers. It does not, however, establish a dedicated authorization framework for consumer AI agents.
For compliance teams, the practical priorities are becoming more specific: determine who authorized an automated interaction, define which actions the authorization covers, verify the identity evidence independently and preserve records of decisions involving elevated risk.
A delegated assistant requesting a routine account update should not automatically receive the same authority as the verified customer approving a withdrawal.
Nor should an automated application be considered fraudulent solely because software participated in its submission.
The next test for crypto identity providers will be whether their systems can verify not only the customer behind an account, but also the scope of authority granted to software acting on that customer’s behalf.
That requirement could reshape onboarding and account security long before AI agents become a routine feature of cryptocurrency trading.
The information presented in this article is intended for informational purposes only and should not be interpreted as financial, investment, or trading advice. Coinspress.com does not promote or advocate for any particular investment strategy, asset, or cryptocurrency project. Cryptocurrency markets are highly volatile and unpredictable – always perform your own research and seek guidance from a qualified financial professional before making any investment decisions.











