0 cumulative citations
View corpus contextAI and blockchain are promising tools to curb fraud across alternative finance platforms, yet the field is held back by a dearth of labeled fraud data and inconsistent evaluation practices, limiting robust evidence on real-world effectiveness.
Taxonomy of fraud types in alternative finance using hybrid systematic review
openalex
review_meta
low evidence
7/10 relevance
Summary only
summary available; pdf_status=paywall
DOI
Source
PDF
A systematic review finds that AI/ML, NLP, and blockchain are the leading technological approaches to detect and mitigate fraud in alternative finance—especially crowdfunding—but progress is hampered by a lack of labeled fraud data and heterogeneous methods.
Citation observations
Cumulative provider counts captured on specific dates; providers are never combined.
0 cumulative citations
View corpus contextAlternative finance platforms, including crowdfunding, peer-to-peer lending, equity-based platforms, and token-based fundraising mechanisms, have become important channels for financing entrepreneurial, social, and investment-oriented initiatives. Yet their reliance on digital intermediation, dispersed participation, and information asymmetry creates opportunities for fraud, undermining trust, investor protection, and platform sustainability. This study provides a systematic review of fraud detection and prevention in alternative finance, with crowdfunding emerging as the most extensively represented empirical domain. Methodologically, the paper combines a PRISMA-guided systematic literature review with a hybrid topic-modeling strategy that integrates neural topic modeling and probabilistic refinement, thereby supporting both transparent corpus selection and data-driven thematic synthesis. The findings show that Artificial Intelligence (AI), Machine Learning (ML), Natural Language Processing (NLP), and blockchain-based mechanisms are recurrently discussed as promising tools for detecting, preventing, or mitigating fraud. AI and ML approaches are mainly used to identify anomalies, suspicious textual patterns, behavioral signals, and transaction irregularities, while blockchain-based approaches are associated with transparency, traceability, smart contracts, and conditional fund release. The review also shows that fraud differs across alternative finance models, ranging from campaign misrepresentation and intentional and premeditated non-delivery in crowdfunding to borrower or platform misreporting in lending-based models and misleading disclosures or white-paper manipulation in ICO/STO contexts. A central challenge across the literature is the scarcity of labeled fraud data, which limits the use and benchmarking of supervised ML models. Overall, this study contributes by linking a reproducible hybrid SLR methodology to a structured synthesis of fraud types, platform-specific vulnerabilities, and AI-, ML-, and blockchain-based detection strategies in alternative finance.
Summary
Main Finding
Alternative finance platforms (crowdfunding, peer-to-peer lending, equity platforms, ICO/STO/token-based fundraising) face recurring fraud risks due to digital intermediation, dispersed participation, and information asymmetry. The literature—dominated empirically by crowdfunding—identifies AI/ML/NLP and blockchain mechanisms as the most promising technological approaches for fraud detection, prevention, and mitigation. A central methodological and practical constraint is the scarcity of labeled fraud data, which limits supervised-model development and benchmarking.
Key Points
- Scope: Alternative finance includes crowdfunding (reward & donation), peer-to-peer lending, equity-based platforms, and token/ICO/STO mechanisms. Fraud modalities differ by model.
- Empirical emphasis: Crowdfunding is the most extensively studied empirical domain in the reviewed literature.
- Fraud types by model:
- Crowdfunding: campaign misrepresentation, deliberate non-delivery, fabricated projects.
- Lending: borrower misreporting, loan-application fraud, platform misreporting or collusion.
- Token/ICO/STO: misleading disclosures, white-paper manipulation, token-sale fraud.
- Prominent technological responses:
- AI / ML: anomaly detection, behavioral-signal modeling, transaction irregularity detection, supervised classifiers (where labeled data exist).
- NLP: analysis of campaign texts, pitch language, white papers, and other narrative signals to flag suspicious textual patterns.
- Blockchain: transparency and traceability of transactions, smart contracts for conditional fund release, immutable audit trails.
- Methodological challenge: a pervasive shortage of labeled fraud instances constrains supervised learning, evaluation, and cross-study comparability.
- Research-method contribution: the study couples a reproducible PRISMA-guided systematic literature review with a hybrid topic-modeling pipeline (neural topic modeling plus probabilistic refinement) to support transparent corpus selection and a data-driven thematic synthesis.
Data & Methods
- Literature selection: PRISMA-guided systematic literature review (transparent inclusion/exclusion and corpus curation).
- Thematic synthesis: hybrid topic-modeling strategy that integrates neural topic models with probabilistic refinement to extract latent themes and structure the qualitative synthesis.
- Evidence base: heterogeneous—empirical papers (particularly on crowdfunding), methodological AI/ML proposals, conceptual papers on blockchain applications.
- Identified methodological limitations across studies: limited labeled datasets for fraud, heterogeneous definitions of fraud, varying evaluation protocols.
Implications for AI Economics
- Market trust and participation: Effective detection/prevention technologies can reduce information asymmetry, raise investor confidence, and lower market frictions—supporting platform growth and deeper capital allocation to startups and social projects.
- Technology–policy complementarity: Blockchain-based transparency and AI-driven monitoring are complementary: blockchain can provide auditable data streams while AI can analyze those streams for anomalies. Regulators and platforms should coordinate standards for on-chain data formats and disclosure to enable automated oversight.
- Research priorities:
- Data infrastructure: incentivize creation and sharing of labeled fraud datasets (anonymized or synthetic) and benchmark tasks to allow reproducible ML evaluation.
- Methods: prioritize semi-supervised, unsupervised, anomaly-detection, transfer learning, and synthetic-data approaches that mitigate labeled-data scarcity.
- Interdisciplinarity: combine technical detection with behavioral and institutional analysis to design preventive platform rules (e.g., staged funding via smart contracts).
- Economic modeling: incorporate detection costs, false-positive/false-negative trade-offs, and strategic agent responses (fraudsters adapting to detectors) into models of platform equilibrium and regulation design.
- Policy design: regulators should balance investor protection with innovation—supporting standards for disclosures, mandatory audit trails, and mechanisms (e.g., escrowed/smart-contracted releases) that reduce the payoff to fraud.
- Platform strategy: platforms can deploy layered defenses—NLP screening of narratives, behavioral analytics of backer/borrower activity, transaction-pattern monitoring, and selective on-chain commitments—to reduce fraud incidence while minimizing friction for legitimate users.
Overall contribution: the study maps the landscape of fraud risks and technological countermeasures across alternative finance, highlights methodological gaps (notably labeled-data scarcity), and provides a reproducible literature-synthesis approach that can support future AI- and economics-focused research and policy design.
Assessment
Paper Typereview_meta
Evidence Strengthlow — The paper synthesizes heterogeneous literature dominated by descriptive and methodological studies (especially on crowdfunding) rather than causal empirical research; many contributions are ML method proposals without field validation, and a pervasive scarcity of labeled fraud incidents limits rigorous evaluation and benchmarking.
Methods Rigorhigh — The review uses a reproducible PRISMA-guided corpus selection and a transparent hybrid topic-modeling pipeline (neural topic models plus probabilistic refinement), which are appropriate and rigorous for systematic literature synthesis, though the underlying studies vary widely in rigor.
SampleA PRISMA-curated literature corpus covering alternative finance modalities (crowdfunding—reward & donation—peer-to-peer lending, equity platforms, ICO/STO/token sales), composed of empirical studies (predominantly crowdfunding), methodological AI/ML proposals (anomaly detection, NLP), and conceptual blockchain applications; applied hybrid topic modeling to extract themes. Notably, labeled fraud-instance datasets are scarce across studies.
Themesgovernance adoption innovation human_ai_collab
GeneralizabilityEvidence is heavily concentrated on crowdfunding; findings may not generalize to lending, equity, or token markets with different fraud modalities., Heterogeneous and inconsistent definitions of 'fraud' across studies reduce comparability and external validity., Scarcity of labeled fraud data limits transferability of supervised ML methods and benchmarking across platforms and contexts., Many studies are methodological proposals or simulations rather than field-validated interventions, limiting real-world applicability., Geographic and platform heterogeneity (regulatory regimes, platform rules, user mixes) are not uniformly represented, constraining global generalizability.
Claims (10)
| Claim | Direction | Outcome | Confidence & Evidence | Details |
|---|---|---|---|---|
| Alternative finance platforms face recurring fraud risks associated with digital intermediation, dispersed participation, and information asymmetry. Ai Safety And Ethics | negative | Fraud risk |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Crowdfunding is the most extensively studied empirical domain in the alternative-finance fraud literature. Other | positive | Empirical research coverage |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Fraud modalities differ across alternative-finance models: crowdfunding commonly involves campaign misrepresentation, deliberate non-delivery, and fabricated projects; lending involves borrower or loan-application misreporting and possible platform collusion; and token fundraising involves misleading disclosures, white-paper manipulation, and token-sale fraud. Ai Safety And Ethics | negative | Fraud modality and incidence risk |
Reading fidelity
high
Study strength
medium
|
not reported
|
| AI and machine-learning methods are identified as promising approaches for detecting, preventing, and mitigating fraud in alternative finance. Ai Safety And Ethics | positive | Fraud detection and prevention |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Natural-language-processing methods can use campaign texts, pitch language, white papers, and related narrative signals to identify suspicious textual patterns. Ai Safety And Ethics | positive | Detection of suspicious textual patterns |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Blockchain mechanisms can support fraud mitigation through transaction transparency and traceability, smart contracts for conditional fund release, and immutable audit trails. Regulatory Compliance | positive | Transaction transparency and fraud mitigation |
Reading fidelity
high
Study strength
low
|
not reported
|
| A pervasive shortage of labeled fraud instances constrains supervised-learning development, model evaluation, and comparability across studies. Training Effectiveness | negative | Availability of labeled fraud data and validity of supervised-model evaluation |
Reading fidelity
high
Study strength
high
|
not reported
|
| The reviewed studies use heterogeneous definitions of fraud and varying evaluation protocols, which limit cross-study comparability. Research Productivity | negative | Cross-study comparability |
Reading fidelity
high
Study strength
medium
|
not reported
|
| The study combines a PRISMA-guided systematic literature review with a hybrid topic-modeling pipeline that integrates neural topic modeling and probabilistic refinement. Research Productivity | positive | Transparency and structure of literature synthesis |
Reading fidelity
high
Study strength
high
|
not reported
|
| Effective fraud-detection and prevention technologies could reduce information asymmetry, increase investor confidence, lower market frictions, and support growth in alternative-finance platforms. Consumer Welfare | positive | Investor confidence, market frictions, and platform growth |
Reading fidelity
high
Study strength
speculative
|
not reported
|