Papers
Topics
Authors
Recent
Search
2000 character limit reached

PEB Separation and State Migration: Unmasking the New Frontiers of DeFi AML Evasion

Published 27 Mar 2026 in cs.CR and q-fin.TR | (2603.26290v1)

Abstract: Transfer-based anti-money laundering (AML) systems monitor token flows through transaction-graph abstractions, implicitly assuming that economically meaningful value migration is sufficiently encoded in transfer-layer connectivity. In this paper, we demonstrate that this assumption, the bedrock of current industrial forensics, fundamentally collapses in composable smart-contract ecosystems. We formalize two structural mechanisms that undermine the completeness of transfer-layer attribution. First, we introduce Principal-Execution-Beneficiary (PEB) separation, where intent originators, transaction executors (e.g., MEV searchers), and ultimate beneficiaries are functionally decoupled. Second, we formalize state-mediated value migration, where economic coupling is enforced through invariant-driven contract state transitions (e.g., AMM reserve rebalancing) rather than explicit transfer continuity. Through a real-world case study of role-separated limit order execution and a constructive cross-pool arbitrage model, we prove that these mechanisms render transfer-layer observation neither attribution-complete nor causally closed. We further argue that simply expanding transfer-layer tracing capabilities fails to resolve the underlying attribution ambiguity inherent in structurally decoupled execution. Under modular composition and open participation markets, these mechanisms are structurally generative, implying that heuristic-based flow tracing has reached a formal observational boundary. We advocate for a paradigm shift toward AML based on execution semantics, focusing on the restitution of economic causality from atomic execution logic and state invariants rather than static graph connectivity.

Authors (3)

Summary

  • The paper demonstrates that observable transfer-layer AML methods are inadequate due to novel DeFi execution semantics, including PEB separation and state-mediated migration.
  • It introduces a formal model capturing role-mediated indeterminacy and state-mediated non-closure, validated by simulations achieving up to 93.5% efficiency in value relocation.
  • The study calls for a paradigm shift toward execution-semantic analysis to restore economic causality and effective beneficiary attribution in decentralized finance AML.

Unmasking Structural Limits of DeFi Anti-Money Laundering via PEB Separation and State-Mediated Migration

Introduction

The paper "PEB Separation and State Migration: Unmasking the New Frontiers of DeFi AML Evasion" (2603.26290) rigorously characterizes the limitations of current transfer-based anti-money laundering (AML) approaches in composable DeFi ecosystems. Prevailing AML methodologies operate under the assumption that value migration is isomorphic to observable transfer-layer graphs, facilitating end-to-end tracing, clustering, and risk attribution through heuristics and graph machine learning. This work formalizes and demonstrates the fundamental incompleteness of transfer-layer abstractions in capturing the underlying economic causality of value migration, arising from composable execution semantics and protocol modularity endemic to modern DeFi.

Transfer-Based AML: Core Assumptions and Emerging Challenges

Current industrial AML systems (e.g., Chainalysis, Elliptic, TRM) rely on the flow continuity, initiator visibility, and endpoint linkage assumptions: that economic value migration can be reconstructed from observable token transfers, that suspicious conversions are directly triggered by high-risk actors, and that recipients at the transfer layer are the ultimate economic beneficiaries. Graph-based forensic tools derive attribution and risk scores through heuristic path-based reasoning in the transfer graph, which regulatory frameworks have subsequently enshrined as the basis for compliance and reporting.

The rapid rise of intent-centric and composable DeFi interactions (e.g., limit-order protocols, aggregator-driven routing, searcher/solver markets, and flash liquidity primitives) systematically undermines these assumptions. Off-chain intent signaling, third-party executor mediation, and state/invariant-based routing decouple the observable flow of assets from the actors' economic intentions and roles.

Principal-Execution-Beneficiary (PEB) Separation

The work formalizes PEB separation: a functional decoupling between three core roles—principals (intent originators), executors (transaction initiators, often MEV searchers or solvers), and beneficiaries (economic recipients). Through detailed case analysis of a 38.2M USDC to 38.1M DAI conversion on Ethereum via 1inch limit orders, Uniswap V3, and Aave flash loans, the paper concretely demonstrates how PEB separation manifests in operational DeFi flows. Here, the high-risk principal does not submit the transaction, the MEV executor acts purely as a filler without being an economic beneficiary, and the beneficiary's receipt of assets is disconnected from the observable swap logic.

Under these execution patterns, direct attribution links between principal and beneficiary are irrecoverable from the transfer graph alone, even with complete on-chain data. Instead, the linkage is structurally mediated by protocol logic and stateful contract transitions invisible to static graph traversal. As additional routing strategies and composable venues proliferate, such structurally separated migration patterns become generative and not protocol-specific edge cases.

State-Mediated Value Migration

The second core mechanism is state-mediated value migration—value transfers enforced through execution-level AMM or protocol state transitions, rather than explicit token transfer continuity. The paper constructs a formal cross-pool arbitrage model (extensively proven and empirically validated) in which economic value is transferred from a principal to a beneficiary via two sequential arbitrage loops across different AMM pools, exploiting only regular constant-product invariant dynamics. In this setting, there is no transfer path in the observable graph that uniquely encodes the principal-beneficiary relationship; the economic coupling emerges purely from coordinated state transitions, and the observable token flows are indistinguishable from typical market arbitrage.

The authors provide zero-fee and mainnet simulation proofs showing efficiency of up to 93.5% for value relocation via such constructions, even with prevailing AMM fees and slippage. Attribution ambiguity remains, even when the principal and operator are not separated, highlighting the sufficiency of state mediation alone for transfer-layer causal non-closure.

Transfer-Layer Incompleteness: Formal Model and Theoretical Implications

The work rigorously defines transfer-layer incompleteness by distinguishing between economic migration (enforced atomic balance updates) and transfer-layer recoverability (existence of a unique, deterministic path in the observable transfer graph encoding this relation). Two formal structural failure modes are articulated:

  • Role-Mediated Indeterminacy: Under PEB separation, neither the principal nor the beneficiary is directly observable as transaction initiator or direct recipient, resulting in irreducible ambiguity.
  • State-Mediated Non-Closure: Even absent PEB separation, when value migration is accomplished through protocol state transitions (e.g., AMM arbitrages or flash liquidity), transfer-layer structures do not encode unique causality.

A central theorem asserts that, in composable smart-contract environments, there exists a structurally generative space of execution strategies for which transfer-layer monitoring is neither attribution-complete nor causally closed. These patterns arise from economically rational protocol usage, requiring no protocol aberrations.

Limitations of Current AML Paradigms and Need for Execution-Semantic Monitoring

The implications are significant. Rule-based and graph-machine-learning AML detection is inherently upper-bounded in composable, modular environments. Pattern mutation and adversarial adaptation (akin to malware polymorphism) render motif-based detection fundamentally incomplete; observable transfer graphs can be arbitrarily restructured by adversaries while maintaining economic semantics.

The paper argues for a paradigm shift toward execution-semantic AML analysis: considering call graph structure, contract state changes, and intent resolution logic as first-class inputs to forensic reasoning. Such approaches, while computationally and conceptually more challenging, are necessary for restoring economic causality and beneficiary attribution. This need is echoed by recent advances in EVM execution trace analysis tools and the literature on semantic-driven monitoring (Mao et al., 31 Dec 2025, Weber et al., 2019), but these tools currently lack the attribution intent modeling required for robust AML.

Empirical Validation and Efficiency Considerations

Through mainnet simulation (Uniswap V2 and SushiSwap, WETH/USDT), the authors show efficient value relocation with empirical loss (due to AMM fees and slippage) of approximately 6.5%. Notably, all on-chain token flows are consistent with benign market behavior, and static tracing tools offer no distinctive identifier for such state-mediated economic migration.

Conclusion

This work establishes that the dominant abstraction underpinning industrial and academic AML—transfer-based observation and graph-based attribution—is formally and generatively insufficient in composable DeFi environments. Both PEB separation and state-mediated value migration represent fundamental structural mechanisms by which adversaries can evade detection and obfuscate economic intent. Theoretical advances in AML will require robust execution modeling, integration of semantic analysis, and frameworks for off-chain intent and settlement logic. These findings delineate a new research agenda for RegTech and transaction forensics in the rapidly composable, modular frontier of Web3 finance.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Sign Up to Generate All Videos Subscribe on YouTube

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Sign Up to Generate

Explain it Like I'm 14

What this paper is about (in simple terms)

This paper looks at how people can move money around in decentralized finance (DeFi) in ways that trick today’s anti-money-laundering (AML) tools. Most AML tools watch “who sent tokens to whom” like a big social network of payments. The authors show that, in modern DeFi, real value can move without leaving a clear “send/receive” trail, so those tools can miss what actually happened. They argue we need a new way to track value that looks at what the smart contracts actually did under the hood, not just the visible transfers.

The big questions the paper asks

Before getting into details, the paper questions three common assumptions that many AML systems make:

  • Flow continuity: If value moves, you can see it as a chain of transfers from wallet to wallet.
  • Initiator visibility: The risky person (the suspect) is the one who sends the transaction.
  • Endpoint linkage: The address that receives tokens is the true final owner or beneficiary.

The paper asks: Are these assumptions still true in DeFi, where many actions are chained together by smart contracts and third parties? And if not, what replaces them?

How the authors studied the problem (ideas and methods made simple)

The authors describe two key ideas that break the “follow the transfers” approach:

  • PEB separation (Principal–Execution–Beneficiary): Think of three roles in a deal:
    • The Principal (P): the person who wants to make the trade (and may be the suspicious one).
    • The Executor (E): a third party (often a bot) who actually sends the on-chain transaction.
    • The Beneficiary (B): the person who ends up with the money.
    • These three can be different people. That means the risky person might never press “send,” and the person who presses “send” might just be a bot, and the receiver might be someone else entirely.
  • State‑mediated value migration: Sometimes value moves because of how a contract’s internal “state” changes, not because one address directly transfers to another. For example, an Automated Market Maker (AMM) is like a vending machine that always keeps its prices balanced. If you push and pull on its prices the right way (arbitrage), you can shift value between people without a clean, direct transfer line connecting them.

To make these ideas concrete, they do two things:

  1. Real case study: They analyze a real Ethereum transaction where a big amount of USDC became DAI through a limit order on 1inch. The important part: the risky wallet signed an off-chain order (like a permission slip), but a separate bot executed the trade on-chain, and a third address got the DAI. There was no simple “P sent to B” payment to follow.
  2. Constructive model: They build a step-by-step example using two AMM pools. By doing a clever sequence of swaps (including flash loans, which are super-short loans paid back in the same transaction), they show value can move from P to B even though there’s no straightforward transfer path linking them. It looks like normal arbitrage on the surface, but it effectively shifts P’s value to B.

In everyday language: Instead of watching who handed cash to whom, you also need to watch what the machines (smart contracts) did inside—how they changed their internal counters and prices—because that’s where the value can “teleport.”

What they found and why it matters

Here are the main takeaways, explained plainly:

  • Following transfers is not enough: In DeFi, the important action can be hidden inside how smart contracts run, not in the visible “A sent tokens to B” edges. That means the transfer graph doesn’t always show who really benefited from whose money.
  • PEB separation breaks the “who pressed send” rule: A risky person can sign an off-chain order, a bot can do the on-chain work, and a separate address can receive the funds. The risky person may never appear as the transaction sender, and the final receiver may never have a direct link back to the risky person.
  • State changes can move value without a clean trail: By shifting prices and reserves across liquidity pools, an operator can make value end up at a chosen wallet without creating an obvious, unique “money trail” between the source and destination. It looks like normal market activity.
  • Making the graph bigger won’t fix it: Even if you trace more hops or use fancier graph tricks, the core problem remains. Different behind-the-scenes executions can produce the same visible transfers. So the graph can’t tell you the true story by itself.

Why this matters: If AML tools only look at transfers, clever bad actors can use DeFi’s building blocks (limit orders, AMMs, MEV bots, flash loans) like Lego pieces to hide who actually benefited from stolen or dirty funds. That weakens current compliance and enforcement.

What this could change going forward

The paper argues for a shift in AML from “transfer tracking” to “execution understanding”:

  • Look at execution semantics: Instead of just mapping transfers, analyze the transaction’s call steps, which contracts called which, and how key state variables (like AMM reserves) changed. In short, move from “who sent what” to “what exactly happened.”
  • Connect off-chain intent to on-chain action: Many orders are signed off-chain and later filled by others. AML needs ways to link those off-chain intents to the on-chain execution and to the final beneficiary.
  • Expect shape-shifting behavior: Just like advanced computer viruses change their appearance, money movements in DeFi can change their visible patterns while doing the same thing. Rules and pattern-matching alone won’t keep up.
  • Policy and tooling impact: Regulators and analytics companies will need new tools that can replay transactions, read internal state changes, and reason about economic cause and effect—so they can tell who really gained from whose losses, even when there’s no simple transfer path.

In short: The paper says DeFi has outgrown AML methods that only follow token transfers. To catch modern laundering tricks, we must understand how the smart contracts’ “machines” actually ran and how that created real economic outcomes, not just what the transfer graph shows.

View Paper Prompt  View All Prompts 

Knowledge Gaps

Knowledge gaps, limitations, and open questions

Below is a consolidated list of what remains missing, uncertain, or unexplored, framed to guide concrete future research.

  • Lack of prevalence measurement: no empirical quantification of how often PEB separation and state-mediated value migration occur across protocols, chains, and time.
  • Missing taxonomy: no systematic categorization of PEB separation patterns (e.g., RFQ, UniswapX, CoW, OTC, brokered fills, solver auctions) and their distinct attribution failure modes.
  • Informal identifiability results: the central “Transfer-Layer Incompleteness” theorem is informal; necessary and sufficient conditions for non-identifiability are not formally proved in an information-theoretic or observability model.
  • Unspecified adversary model: no explicit assumptions about attacker capabilities, costs, coordination with searchers, access to flash liquidity, or willingness to accept execution risk.
  • Execution-semantic AML blueprint absent: no end-to-end system design for execution-semantic monitoring (data ingestion, trace normalization, state-diff extraction, invariant inference, attribution logic).
  • Scalability unanswered: no complexity analysis or engineering plan for running call-graph/state-invariant analytics at chain scale and across multiple chains/L2s.
  • Ground truth deficit: no methodology to build labeled datasets for “economic causality” and “beneficiary” attribution needed to train/evaluate execution-semantic AML.
  • Ambiguity resolution methods missing: no probabilistic/causal inference framework to resolve multiple plausible beneficiary linkages (e.g., Bayesian priors, SCMs, counterfactual replays).
  • Cross-domain scope limited: analysis is intra-chain; no treatment of bridges, rollups, intent-centric L2s, cross-chain MEV, or privacy-preserving chains and their impact on identifiability.
  • PBS/private orderflow opacity unquantified: no study of how builder markets, private mempools, orderflow auctions, or encryption affect observability and AML coverage.
  • Non-atomic laundering unexplored: patterns that spread across blocks (e.g., non-atomic arbitrage sequences, time-separated state migration) are not modeled or measured.
  • Frictions as detection levers unstudied: no theoretical bounds showing how fees, slippage, and liquidity constraints limit relocation efficiency or create detectable signatures.
  • Discriminative signal design absent: no proposed feature set from execution traces (e.g., invariant deltas, callback motifs, allowance usage, flash-loan/flash-swap patterns) to distinguish laundering from benign arbitrage.
  • Protocol-level mitigations not explored: no design proposals (e.g., proof-carrying orders, on-chain intent attestations, solver identity attestations, receiver constraints) to restore attribution without sacrificing composability.
  • Regulatory integration unclear: no pathway to incorporate execution-semantic evidence into VASP workflows, STRs, or Travel Rule compliance; evidentiary standards and auditability not addressed.
  • Privacy and governance implications omitted: no discussion of user privacy risks, data retention, or governance for deep trace/state analysis.
  • False positive control open: no methodology to reduce confusion between normal market-making/arbitrage and laundering under execution-level analysis.
  • Limited protocol coverage: only a 1inch limit-order case is analyzed; UniswapX, CoW Protocol, off-chain RFQ brokers, and other solvers are not empirically evaluated.
  • AMM generality restricted: formal construction targets constant-product CFMMs; concentrated liquidity (e.g., Uniswap V3), stableswap, hybrid CFMMs, and fee-on models lack formal analysis.
  • Sensitivity analysis missing: no exploration of how fee tiers, liquidity depth, volatility, and pool correlation affect relocation feasibility and detectability.
  • Cost-benefit for adversaries unmodeled: no economic model comparing these evasion tactics with alternatives (mixers, bridges, OTC cash-outs), including risk and cost trade-offs.
  • Data requirements under-specified: required observables (execution traces, storage diffs, decoded calldata, solver logs) and their availability, standardization, and reproducibility are not detailed.
  • Definition gaps: “economic causality” and “beneficiary” are not fully formalized (e.g., handling shared profits, solver rebates, multi-recipient settlements).
  • Minimal observability question open: no characterization of the minimal additional signals needed to restore identifiability (e.g., publication of signed order digests, solver attestations, receipt commitments).
  • Decidability and complexity open: no results on the decidability/complexity of inferring economic causality from EVM traces and state transitions for general contract classes.
  • Benchmarking absent: no benchmark datasets, tasks, or evaluation metrics for execution-semantic AML; no head-to-head comparison with TxSpector, MFTracer, or commercial tools on PEB/state-mediated cases.
  • Reproducibility limited: appendices lack artifacts (code, configs, seeds, transaction bundles) to reproduce simulations and proofs under varying market conditions.
  • Benign confounders not cataloged: no catalog of benign execution motifs that mimic laundering patterns to guide negative controls in model training.
  • Lifecycle and game-theory unaddressed: no dynamic/adaptive (metamorphic) evasion model or game-theoretic analysis of AML–adversary co-evolution.
  • Chain vantage limitations unstudied: feasibility and utility of mempool-level vs. post-state vantage, including access to private orderflow logs, remains unexplored.
  • Oracle and pricing effects ignored: no analysis of oracle-mediated price moves, manipulation, or cross-venue latency as signals or confounders for AML intent.
  • Post-trade integration open: no methods to fuse execution-semantic signals with off-chain KYC/CEX data to recover attribution despite on-chain ambiguity.
  • ZK/cryptographic aids unexplored: no discussion of privacy-preserving attestations (e.g., ZK proofs of order–beneficiary linkage) that could support AML without revealing full identities.
View Paper Prompt  View All Prompts 

Practical Applications

Below is an overview of practical applications derived from the paper’s findings on PEB separation and state‑mediated value migration, grouped by deployment horizon. Each item includes target sectors, likely tools/products/workflows, and key assumptions or dependencies that affect feasibility.

Immediate Applications

  • Execution-aware AML analytics for DeFi
    • Sectors: Finance (RegTech, compliance), Software (blockchain analytics), Cybersecurity
    • What: Augment existing transaction-tracing platforms with execution-semantic analysis (EVM trace replay, call-graph inspection, invariant/state-delta reasoning) to detect PEB triads and state-mediated migrations that evade transfer-only tracing.
    • Tools/products/workflows:
    • “Execution-semantic AML engine” that ingests full EVM traces and reconstructs economic causality from call graphs and AMM reserve changes.
    • Rule packs for intent protocols (e.g., 1inch, UniswapX) to identify off-chain signed orders, solver/filler roles, and receiver decoupling.
    • “PEB triad detector” and “State-mediated migration detector” modules integrated into risk scoring.
    • Assumptions/dependencies: Availability of high-fidelity traces (including internal calls/events); compute/storage capacity for replay at scale; protocol-specific decoders; acceptable false-positive rates; access to reliable entity labels.
  • Executor-centric investigative workflows for law enforcement and forensics
    • Sectors: Finance (law enforcement, investigations), Policy (supervision), Software (forensic tooling)
    • What: Shift from source-to-destination transfer tracing to “executor-first” attribution by clustering searchers/solvers/builders and linking them to beneficiary addresses.
    • Tools/products/workflows:
    • Playbooks to pivot on solver/filler EOAs/contracts, builder/relay identifiers, and settlement contracts.
    • Subpoena-ready data extracts linking off-chain order books (EIP-712 signatures) to on-chain fills.
    • Builder/relay log correlation where cooperative.
    • Assumptions/dependencies: Cooperation from aggs/solvers/builders; preserved logs; discovery/legal powers; off-chain data access; jurisdictional reach.
  • Risk-scoring features for role separation and state-mediated flows
    • Sectors: Finance (exchanges/VASPs, custodians, insurance), RegTech
    • What: Introduce “execution-latent risk” indicators when funds arrive via third-party fills, receivers differ from makers, or inflows coincide with AMM state patterns indicative of relocation.
    • Tools/products/workflows:
    • Feature engineering: executor-initiated settlement flags, maker/receiver mismatch, flash-swap/flash-loan involvement, anomalous AMM reserve shifts.
    • Adjusted case-management triggers and enhanced due diligence (EDD) rules.
    • Assumptions/dependencies: Access to protocol event schemas; platform willingness to accept nuanced, probabilistic indicators; model calibration to minimize benign arbitrage false alarms.
  • Protocol audit checks for AML observability
    • Sectors: Software (security audits), Finance (DeFi protocol teams)
    • What: Extend audit scope to include “AML observability” findings (e.g., receiver decoupling, arbitrary predicate callbacks) and recommendations for transparency.
    • Tools/products/workflows:
    • Audit checklists for order settlement contracts to ensure optional maker-receiver linkage events and semantic breadcrumbs.
    • Static+dynamic analysis to flag opaque settlement pathways.
    • Assumptions/dependencies: Protocol teams accept recommendations; minimal performance overhead for additional events/logging.
  • VASP deposit screening beyond transfer lineage
    • Sectors: Finance (exchanges/custodians), RegTech
    • What: Enrich deposit provenance checks with execution context (e.g., funds sourced from DEX settlement flows with PEB patterns) to trigger EDD.
    • Tools/products/workflows:
    • Ingestion of on-chain settlement events and solver addresses into compliance rules.
    • Automated case notes linking deposit TXs to execution semantics rather than simple address taint.
    • Assumptions/dependencies: Stable event schemas across aggregators; access to chain history; operational tolerance for additional review volume.
  • MEV/builder/searcher internal compliance policies
    • Sectors: Finance (MEV infrastructure), Policy (self-regulation)
    • What: Voluntary policies and logging for solvers/builders (e.g., job provenance retention, customer acceptance policies) to facilitate post-incident attribution.
    • Tools/products/workflows:
    • Internal registries of fills with maker/receiver metadata; audit trails aligned with privacy and competition constraints.
    • Assumptions/dependencies: Commercial incentives to self-regulate; competitive neutrality; regional legal expectations.
  • Academic datasets and benchmarks for execution-semantic AML
    • Sectors: Academia, Software (open-source)
    • What: Curate labeled corpora of PEB-separated settlements and state-mediated migrations; provide trace-level ground truth and baseline detectors.
    • Tools/products/workflows:
    • Mainnet-fork reproductions; trace archives; shared evaluation metrics for attribution completeness, not only transfer accuracy.
    • Assumptions/dependencies: Safe disclosure practices; reproducibility; long-term hosting for large trace data.
  • Execution-semantic query and rule languages
    • Sectors: Software (dev tools), Academia
    • What: Lightweight DSLs to express “economic causality” queries over traces (e.g., Datalog-like rules on call trees and state diffs).
    • Tools/products/workflows:
    • Libraries to compute AMM invariant deltas and identify coordinated state transitions across pools; prebuilt rules for popular DEXes.
    • Assumptions/dependencies: Standardized trace formats; maintainable protocol adapters; developer adoption.
  • Wallet UX safeguards for intent signing
    • Sectors: Daily life (end users), Software (wallets)
    • What: Surface the designated receiver and executor constraints in off-chain order signing; warn on receiver decoupling and broadly-fillable predicates.
    • Tools/products/workflows:
    • Signing UIs that parse EIP-712 orders, highlight receiver mismatch, and offer “no decoupled receiver” toggles by default.
    • Assumptions/dependencies: Wallets can decode evolving intent schemas; users respond to warnings; minimal friction for legitimate use.

Long-Term Applications

  • End-to-end execution-semantic AML platforms
    • Sectors: Finance (enterprise RegTech), Software (infrastructure)
    • What: Real-time systems that reconstruct economic causality across L1/L2s, bridges, and intent layers, unifying traces, off-chain orders, and builder logs to produce attribution beyond transfer graphs.
    • Tools/products/workflows:
    • Stream processors at builder/relay layers; cross-domain call-graph stitchers; causal inference engines; confidence-scored beneficiary attribution.
    • Assumptions/dependencies: Access to multi-domain telemetry (privacy/security constraints); standard APIs; significant compute budgets; industry buy-in.
  • Standards for AML observability in intent-based protocols
    • Sectors: Finance (DeFi), Policy (industry consortia/standards bodies)
    • What: “Intent Travel Rule”–like standards mandating or recommending observability artifacts (e.g., maker/receiver commitments, settlement-link events) compatible with privacy.
    • Tools/products/workflows:
    • EIP proposals for canonical events; interoperability profiles for aggregators/solvers; compliance attestations.
    • Assumptions/dependencies: Community governance approval; harmonization with privacy expectations; backward-compatibility pathways.
  • Privacy-preserving attestations of execution causality
    • Sectors: Finance (compliance tech), Cryptography
    • What: Zero-knowledge attestations that a settlement corresponds to a valid maker-receiver relation or policy-compliant intent without revealing full details.
    • Tools/products/workflows:
    • ZK circuits for order provenance; on-chain verifiers; VASP APIs to request/verify attestations on deposits.
    • Assumptions/dependencies: Efficient ZK primitives; protocol integration; incentive alignment for attestations.
  • Regulatory frameworks for open execution markets
    • Sectors: Policy/regulation
    • What: Guidance and/or licensing for solver/builders as regulated service providers; recordkeeping and reporting expectations tailored to composable DeFi.
    • Tools/products/workflows:
    • Tiered obligations (e.g., size- or activity-based); cross-border information-sharing mechanisms for solver networks.
    • Assumptions/dependencies: Legal clarity on roles and obligations; international coordination; enforcement practicality.
  • Cross-chain and cross-intent-domain semantic monitoring
    • Sectors: Finance (multi-chain DeFi), Software (data platforms)
    • What: Attribution across rollups, appchains, bridges, and intent networks where execution causality spans domains.
    • Tools/products/workflows:
    • Cross-domain identity resolution; time-synchronized trace alignment; bridge event semantics; multi-ledger causal graphs.
    • Assumptions/dependencies: Standardized bridge/intents telemetry; resolver services; complex operations/SRE.
  • Formal verification for AML-relevant observability properties
    • Sectors: Software (formal methods), Security audits
    • What: Prove that contracts either emit sufficient observability signals for AML or enforce bounds on receiver decoupling and predicate complexity.
    • Tools/products/workflows:
    • Property templates (e.g., “beneficiary binding”); SMT-based tools and proof-carrying code artifacts included in audits.
    • Assumptions/dependencies: Mature executable semantics and tooling; developer training; balancing flexibility vs. constraints.
  • ML for economic intent inference from traces
    • Sectors: Academia, RegTech
    • What: Learning models that infer principal-beneficiary ties and laundering likelihood from call graphs, solver behavior, and state transitions, beyond path motifs.
    • Tools/products/workflows:
    • Graph neural nets over execution DAGs; simulation environments for synthetic training data; active learning with analyst feedback.
    • Assumptions/dependencies: High-quality labels; robustness to distribution shift; interpretability in compliance contexts.
  • Protocol-level risk-adaptive controls
    • Sectors: Finance (DeFi protocol design)
    • What: DEXes and aggregators implement optional constraints or dynamic fees for high-opacity patterns (receiver decoupling, unbounded predicates), or require attestations for certain order types.
    • Tools/products/workflows:
    • Risk-adjusted fee schedules; allowlists/registries for verified solvers; opt-in compliance modes for institutions.
    • Assumptions/dependencies: Market acceptance; avoiding anti-competitive effects; governance alignment.
  • User-level guardians integrated with intent markets
    • Sectors: Daily life (retail users), Software (wallets/intents)
    • What: Smart guardians that co-sign or vet off-chain intents against policy profiles (e.g., block receiver decoupling or unknown executors), with recoverability hooks.
    • Tools/products/workflows:
    • Policy engines in wallets; social recovery co-signers; integration with solver networks for policy-respecting fills.
    • Assumptions/dependencies: UX that users accept; ecosystem support for policy-aware solvers; minimal latency overhead.
  • Testbeds and sandboxes for AML research on composability
    • Sectors: Academia, Public–private partnerships
    • What: Controlled environments with configurable role separation and AMM parameters to evaluate detectors/controls before mainnet deployment.
    • Tools/products/workflows:
    • Public datasets, reproducible forks, scenario catalogs; red-team/blue-team exercises.
    • Assumptions/dependencies: Sustained funding; participation from protocol teams and analytics vendors.

These applications collectively move AML practice from transfer-graph heuristics toward execution-semantic causality. Near-term wins come from reusing existing trace and MEV analysis tooling, while longer-term impact requires standardization, cryptographic attestations, and regulatory adaptation to the realities of composable execution.

View Paper Prompt  View All Prompts 

Glossary

  • Allowance (ERC-20): An approval that lets a smart contract or another address transfer a token holder’s assets up to a specified limit. "which is sourced from P via an allowance."
  • Arbitrage: A strategy that exploits price differences across markets or pools to extract profit or rebalance prices. "a constructive cross-pool arbitrage model"
  • Atomic transaction: An all-or-nothing on-chain execution where all steps either complete or revert together within a single transaction. "the economic migration is deterministically enforced within a single atomic transaction."
  • Automated Market Maker (AMM): A smart-contract-based exchange that prices assets via reserve-based formulas instead of order books. "a value migration structure implemented within a constant-product Automated Market Maker (AMM) system"
  • Call graph: The execution-level network of function calls across contracts within a transaction. "requiring explicit reasoning about call graphs and internal state transitions"
  • Constant-product AMM: An AMM variant that maintains the invariant x·y=k for its asset reserves. "a value migration structure implemented within a constant-product Automated Market Maker (AMM) system"
  • Cross-pool arbitrage: Arbitrage conducted between two or more liquidity pools trading the same pair. "a constructive cross-pool arbitrage model"
  • Customer Due Diligence (CDD): Regulatory checks to verify customers’ identities and assess risks. "requiring customer due diligence (CDD)"
  • DeFi (Decentralized Finance): Permissionless financial services built on smart contracts and public blockchains. "Composable decentralized finance (DeFi)"
  • Economic migration: Execution-enforced movement of economic value from one address to another within an atomic structure. "An execution instance induces an economic migration of asset AA from PP to BB"
  • EIP-712: An Ethereum standard for typed structured data signing to produce verifiable off-chain signatures. "created and signed off-chain as EIP-712 structured data"
  • Entity clustering: Grouping blockchain addresses into inferred real-world or organizational entities based on behavioral heuristics. "entity clustering, labeling, and risk scoring"
  • Externally Owned Account (EOA): A user-controlled Ethereum account operated by a private key (as opposed to a smart contract). "delivered to a distinct EOA."
  • Execution semantics: The behavior and meaning of program execution, including state transitions and effects beyond transfer edges. "We advocate for a paradigm shift toward AML based on execution semantics"
  • Execution-trace analysis: Inspecting low-level execution steps and internal calls to infer behaviors and flows. "MFTracer introduces high-fidelity execution-trace analysis to resolve illicit money flows concealed within complex internal calls"
  • FATF Travel Rule: A requirement for VASPs to transmit originator and beneficiary information alongside transfers. "the implementation of the FATF Travel Rule for information exchange"
  • Flash loan: An uncollateralized loan that must be borrowed and repaid within the same transaction. "borrows DAI via Aave flash liquidity."
  • Flash swap: An AMM feature allowing assets to be borrowed from a pool and repaid within a callback in the same transaction. "Variant using AMM flash swap instead of external flash loan."
  • Graph Convolutional Networks (GCNs): Neural networks that operate on graph-structured data for tasks like classification. "Weber et al. applied GCNs for illicit transaction classification"
  • Intent-centric architecture: A design where users express off-chain signed intents/orders that independent executors fulfill on-chain. "In this intent-centric architecture"
  • Limit order: An order to buy or sell at a specified price or better, often signed off-chain and settled by third parties. "via a 1inch limit order"
  • Maximal Extractable Value (MEV): The value that can be captured by reordering, inserting, or censoring transactions in a block’s execution. "MEV incentives ensure that signed intents are executed efficiently"
  • Mempool: The pool of pending transactions broadcast to the network prior to inclusion in a block. "the economic intent never appears in the mempool"
  • msg.sender: A Solidity variable that holds the immediate caller address of the current function. "the executor is the EOA msg.sendermsg.sender, not the actual illicit actor."
  • PEB separation (Principal–Execution–Beneficiary separation): Functional decoupling of who originates intent, who executes it, and who ultimately benefits. "We formalize PEB separation in the AML context"
  • Predicate (smart contracts): A boolean condition that must hold for execution to proceed, often used to guard orders or callbacks. "Predicate conditions and pre/post-interaction callbacks allow arbitrary embedded logic"
  • RegTech: Technology that supports regulatory compliance processes and supervision. "\keywords{Anti-Money Laundering \and DeFi \and PEB Separation \and Transaction Tracing \and RegTech.}"
  • State-mediated value migration: Movement of value enforced by contract state transitions/invariants rather than explicit transfer paths. "we formalize state-mediated value migration"
  • Suspicious Transaction Reporting (STR): Mandatory reporting of transactions suspected to involve illicit activity. "suspicious transaction reporting (STR)"
  • Taint analysis: Methods that propagate risk or “contamination” through transaction graphs to trace illicit flows. "taint analysis"
  • Transfer graph: A directed graph whose edges are token transfer events between addresses. "Let G=(V,E)G = (V, E) denote the transfer graph"
  • Transfer-layer abstraction: Modeling monitoring solely via observed transfers and their connectivity, abstracting away execution details. "Transfer-layer abstraction is not causally closed with respect to economic migration"
  • Virtual Asset Service Provider (VASP): A regulated intermediary that facilitates virtual asset activities like exchange or transfer. "Virtual Asset Service Providers (VASPs)"
View Paper Prompt  View All Prompts 

Collections

Sign up for free to add this paper to one or more collections.

Sign Up

Tweets

Sign up for free to view the 3 tweets with 73 likes about this paper.

Sign Up for Free