Research
Bridging the Cybersecurity Gap Between Web2 and Web3 -- An Incident-Based Analysis of Organizational and Application-Level Security Failures
Most Web3 security research looks at smart contracts and blockchain protocols, but the authors argue that a large share of real-world losses actually comes from off-chain systems, organizational…

- arXiv
- 2605.18484
- Published
- 2026-05-18
- Authors
- Tarkan Yavas, Arslan Brömme
AI summary
Overview
Research area: Blockchain and Web3 security governance — specifically the organizational, operational and application-level failure modes behind high-impact crypto incidents, and how they map onto established Web2 security frameworks.
Technical level: Intermediate. Readers should be comfortable with concepts like OWASP vulnerability categories, ISMS/ISO/IEC 27001 control domains, private key custody and validator infrastructure. No cryptography or code-level knowledge is required.
Scope (one sentence): A qualitative, three-case incident analysis that maps the Bybit (2025), Ronin Network (2022) and DMM Bitcoin (2024) breaches onto OWASP Top 10:2021 categories and governance domains, and derives 25 blockchain-specific control categories (BCMS-01 to BCMS-25) intended to operationalize ISMS frameworks for Web3 organizations.
What This Paper Is About
Most Web3 security research looks at smart contracts and blockchain protocols, but the authors argue that a large share of real-world losses actually comes from off-chain systems, organizational processes and human workflows. The paper takes three of the highest-damage publicly documented Web3 incidents and asks how far they reflect failures of well-known Web2 security practice, and which control areas generic frameworks fail to cover. From that mapping it proposes a blockchain-specific control catalogue that plugs into existing information security management systems rather than replacing them.
Key Contributions
-
An empirical characterization of off-chain failure patterns. The authors reconstruct the primary attack vector and secondary enabling conditions for three high-impact incidents, showing that losses occurred without any compromise of blockchain cryptographic primitives or consensus.
-
A structured mapping to OWASP Top 10:2021. Each incident is classified into OWASP vulnerability categories with a primary/secondary relevance rating, producing a cross-case comparison table (Table 1) that identifies which categories recur and which do not apply at all.
-
A mapping to organizational control domains. The same incidents are mapped to high-level ISMS/governance domains such as access governance, key and credential management, third-party risk and operational monitoring.
-
A derived, prioritized set of blockchain-specific control categories. The paper presents 25 BCMS controls organized into 12 thematic groups, motivated and prioritized by the incident evidence, and states that the catalogue is based on the Blockchain Cybersecurity Controls framework developed by the authors' own organization (Chain Horizon) and is available in a GRC tool.
Main Findings
-
The incidents were not blockchain-protocol failures. In all three cases the underlying chain cryptography held; attackers instead compromised third-party signing frontends, validator key material, or custodial private keys.
-
Bybit (21 February 2025, approximately $1.5 billion USD). Interface manipulation and social engineering via a compromised third-party wallet frontend (Safe{Wallet}) caused legitimate signers to approve malicious transactions. Mapped primarily to A08 (Software and Data Integrity Failures), A06 (Vulnerable and Outdated Components), A04 (Insecure Design) and A09 (Security Logging and Monitoring Failures).
-
Ronin Network (March 2022, approximately $615 million USD: 173,600 ETH plus 25.5 million USDC). Compromise of validator private key material and operational control, with insufficient segregation of validator infrastructure. Mapped primarily to A01, A04, A05 and A07. Laundering activity was linked to DPRK cyber threat actors by the U.S. Department of the Treasury, with independent blockchain forensics corroborating the linkage.
-
DMM Bitcoin (May 2024, more than $300 million USD, approximately 4,500 BTC). Compromise of private keys for custodial wallets, reflecting failures in key lifecycle management and privileged access control. Mapped primarily to A02 (Cryptographic Failures) and A01.
-
No single OWASP category dominates across all three cases. A04 (Insecure Design) is of primary relevance in two of the three incidents, suggesting architectural and governance design decisions—rather than implementation bugs—are the recurring enabler. A03 (Injection) and A10 (Server-Side Request Forgery) were not applicable to any case.
-
Three BCMS controls are of primary relevance in all three incidents: BCMS-01 (Blockchain-specific threat analysis and risk management), BCMS-16 (Access controls for blockchain administrators) and BCMS-22 (Blockchain governance and compliance). The authors treat these as the most critical baseline requirements regardless of operational model or attack vector.
-
Several controls were never implicated: BCMS-08 (Secure coding for smart contracts), BCMS-18 (Use of zero-knowledge proofs) and BCMS-23 (Protection against 51% attacks) were not directly implicated in any analysed case, consistent with the finding that dominant failures sit outside the protocol layer. (Table 2 also shows no relevance for BCMS-24.)
-
A shared threat-actor profile emerged rather than being designed in. All three cases converged on DPRK-attributed, state-sponsored actors; the authors state this was not a selection criterion, and it limits generalizability to opportunistic attackers or malicious insiders.
-
Physical coercion is flagged as a separate emerging threat category. The discussion cites cases including the 2025 abduction in France of a major hardware-wallet company's co-founder and his partner, another French kidnapping case linked to a wealthy cryptocurrency entrepreneur's family, and a U.S. case of an investor allegedly held captive to coerce disclosure of private keys. These bypass technical controls entirely.
Methodology in Plain English
This is a qualitative, exploratory study—no statistics, no benchmark, no dataset of incidents beyond the three selected cases. The authors drew only on public material: technical post-mortems, official statements by the affected organizations, cybersecurity vendor reports, investigative journalism and law-enforcement advisories.
Cases were chosen purposively, not randomly. The inclusion rules were that the organization primarily operates in Web3, the incident caused a substantial publicly reported financial loss or severe operational disruption, enough public detail existed to reconstruct the attack vector, and the root cause extended beyond smart contract or protocol flaws into off-chain components, workflows, governance or human processes. The authors state they picked the three highest-damage incidents in the Web3 space at the time of writing, covering three different organizational classes: a centralized exchange, a cross-chain bridge, and a custodial service provider.
For each case they then did four things: identified the primary attack vector from public documentation, extracted secondary enabling factors (missing controls, weak segregation of duties, limited monitoring), classified technical weaknesses against OWASP-style categories, and classified organizational weaknesses against ISMS/governance domains. Both authors coded independently and reconciled through structured discussion—a four-eyes principle to reduce confirmatory bias. Ambiguous public information was handled conservatively, with speculative assumptions avoided.
The authors are explicit that this supports analytic, not statistical, generalization: the goal is to inform control design, not to estimate how often any vulnerability occurs across the ecosystem.
Why This Matters
Impact on research. The paper redirects attention from on-chain code to the operational envelope around it. It also contributes a reusable, testable control structure that other researchers can apply to different incident sets, and it surfaces physical coercion as a threat category that conventional cybersecurity controls cannot address.
Real-world applications:
- Exchange and custodian security architecture: the BCMS categories give teams concrete areas to audit—key lifecycle, signer infrastructure, third-party wallet tooling, approval workflow monitoring.
- ISMS implementation and certification scoping: organizations pursuing ISO/IEC 27001 can use the catalogue to extend technology-agnostic controls with blockchain-specific objectives.
- Third-party and vendor assurance: the Bybit case shows how a compromised external signing frontend becomes a critical dependency, arguing for formal integrity assurance and contractual controls over wallet and signing vendors.
- Incident response and detection engineering: the recurring A09 findings point to monitoring gaps around signing workflows and anomalous approval patterns, not just on-chain transaction monitoring.
Industry relevance. The paper targets a governance gap: Web3 firms often operate in fast-moving, innovation-driven environments where the audit and control structures standard in regulated digital infrastructure are not yet systematically established. Because blockchain transactions are irreversible, procedural weaknesses translate into immediate, unrecoverable losses—amplifying the cost of gaps that would be survivable in traditional finance.
Future Directions
- Test the catalogue against other incident datasets. The authors call for evaluating the proposed control categories beyond these three cases, which were selected for financial impact and share a single threat-actor profile.
- Develop measurable control objectives and maturity criteria. The BCMS entries are currently categories with descriptions, not auditable metrics.
- Validate effectiveness empirically in operational Web3 environments—i.e., does adopting these controls actually reduce incident severity?
- Extend the threat model beyond state-sponsored attackers to opportunistic attackers and malicious insiders, and integrate cyber-physical security (coercion, extortion, key-holder personal security) into Web3 risk management.
Target Audience
Security architects and CISOs at exchanges, custodians and blockchain infrastructure providers; GRC and compliance professionals adapting ISO/IEC 27001 or NIST CSF 2.0 to digital-asset environments; auditors assessing Web3 vendors; and academic researchers working on organizational and human-factor security in blockchain systems. It is less suited to readers seeking smart contract vulnerability analysis or quantitative incident statistics—neither is provided.
Authors’ abstract
The rapid adoption of Web3 infrastructures has led to a growing number of security incidents affecting cryptocurrency exchanges, custody services and blockchain-based platforms. While existing research predominantly focuses on vulnerabilities in smart contracts and blockchain protocols, a substantial portion of real-world losses originates from off-chain systems, organizational processes and human-centered operational workflows. This paper presents a qualitative,…
The full abstract on arXiv