Wulf A. Kaal

How can we Best Monitor AI Agents

Full text for verification

How can we Best Monitor AI Agents

Canonical record: https://ssrn.com/abstract=5245185

40 protected claims are extracted from this work.

Source extraction SHA-256: 935c123d02e511201d48ce5801201b47415066b36bbea18d45599e72408e5be1

MDPI

future internet

_Article_

# **How can we Best Monitor AI Agents?**

**Wulf Kaal, Ph.D.**1

School of Law, University of St. Thomas, St Paul, MN 55105, USA

**Abstract:** This paper examines the critical challenge of monitoring AI agent transaction execution within decentralized digital ecosystems, highlighting the deficiencies of traditional centralized AI-driven supervision, including opacity, bias, and systemic vulnerabilities. In response, it proposes a web3 Decentralized Autonomous Organization (DAO)-centric governance model that integrates blockchain technology, federated communication platforms, smart contracts, and Weighted Directed Acyclic Graphs (WDAGs) to deliver an alternative oversight framework. The proposed system ensures unparalleled transparency and accountability through blockchain’s immutable ledger, while decentralized decision-making via community consensus mitigates bias and single points of failure. Federated platforms enhance scalability and privacy by distributing data processing, and smart contracts automate real-time compliance, bolstered by WDAGs’ adaptive governance structure. Validation pools and reputation tokens further empower stakeholders, fostering a dynamic, inclusive monitoring process.

By incorporating feedback loops, this model anticipates and adapts to AI evolution, overcoming scalability, interoperability, and regulatory gaps inherent in existing frameworks. This decentralized approach not only addresses current shortcomings but also establishes a forward-looking standard for secure, compliant, and efficient AI agent management in modern infrastructures.

**Keywords:** Web3; Artificial Intelligence; Blockchain; Decentralization; Data Sovereignty; Privacy; Smart Contracts; Tokenization; Weighted Directed Acyclic Graph; Web3 Governance; DAO;

## **1. Introduction**

The rapid proliferation of artificial intelligence (AI) has ushered in a new era of autonomous digital agents tasked with executing complex transactions across diverse platforms. As these AI agents become increasingly integral to financial, logistical, and operational systems, effective monitoring of their transaction execution emerges as a critical challenge. Traditional, centralized AI-driven supervision methods are often hindered by limited transparency, susceptibility to bias, and the risk of single points of failure. Consequently, there is a growing imperative to explore alternative oversight frameworks to ensure security, compliance, and efficiency in managing AI agent transactions.

In response to these challenges, this paper proposes a decentralized governance system that integrates web3 community software with federated communication platforms. This decentralized mechanism guarantees complete transparency and public accountability, thereby enabling independent verification of AI agent activities.2 Moreover, by leveraging the distributed nature of web3 technology, the proposed framework facilitates consensus-driven decision-making among a diverse stakeholder

https://doi.org/10.3390/xxxxx

_Future Internet_ **2025** , _17_ , x

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW2 of 21

community. This decentralized approach minimizes systemic biases and reduces the risk associated with centralized oversight, ensuring a more resilient and aligned monitoring process.3

Furthermore, the integration of federated communication platforms, such as Matrix with its Synapse server, distributes data processing tasks across multiple nodes. This federated model not only scales the monitoring process efficiently but also enhances data privacy by processing sensitive information locally rather than in a single centralized repository.4 In addition, the application of smart contracts (self-executing agreements) automates compliance checks and performance validations, ensuring that governance protocols are executed consistently and in real time. Coupled with the structured oversight provided by Weighted Directed Acyclic Graphs (WDAGs), these mechanisms enable dynamic adaptation to emerging standards.5

This paper contends that the proposed decentralized governance system not only overcomes the inherent limitations of centralized, AI-only monitoring approaches but also establishes a robust, transparent, and adaptive framework for overseeing AI agent transactions. By aligning technological oversight with community values and regulatory mandates, this model sets new standards for secure, compliant, and efficient AI system management in modern digital infrastructures.

# **2. Convergence of AI and Blockchain Technology**

The integration of AI with blockchain technology represents a significant development for the digital asset industry, offering robust solutions to persistent challenges in data management, security, privacy, and operational efficiency. This confluence leverages the complementary strengths of both technologies, thereby introducing innovative paradigms for operational frameworks and economic structures.6

This integration transcends a mere technological amalgamation, signifying a profound shift towards a digital ecosystem characterized by heightened security, transparency, and democratization. It may revolutionize industries by enhancing data management practices, fortifying privacy safeguards, advancing decentralization, and fostering novel economic models. Nonetheless, the trajectory towards a fully decentralized AI landscape remains an evolving process, necessitating persistent research, regulatory evolution, and interdisciplinary collaboration to address extant limitations and fully realize its transformative potential.7

Blockchain's immutable ledger functionality provides a critical advantage for AI by ensuring the authenticity and traceability of training data, an aspect particularly vital for systems employing LLMs, where data integrity directly impacts output reliability. Blockchain establishes a verifiable record of data provenance and alterations, essential for sustaining trust and mitigating risks such as data poisoning, thereby ensuring AI models are trained on genuine datasets.8

The decentralization inherent in blockchain technology dismantles traditional barriers posed by data silos and centralized governance, offering significant benefits to small and medium-sized enterprises (SMEs). These entities can now engage in secure, resource-efficient cross-chain transactions, leveraging the resilient and transparent environment blockchain provides. This decentralization fosters collaborative models that enhance operational efficiency and security, while AI algorithms capitalize on blockchain’s real-time data processing capabilities to enable dynamic, adaptive systems.9

For privacy and security, the integration of AI with decentralized physical infrastructure networks (DePINs) is a significant advancement. Blockchain’s cryptographic mechanisms ensure data privacy and verifiability, critical for AI operations involving sensitive information. DePIN decentralizes control over physical devices, enabling secure AI computations without compromising user privacy.10

Blockchain facilitates the democratization of AI services through tokenization and decentralized marketplaces, managing access to AI resources transparently and securely

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW3 of 21

via blockchain-based tokens. This approach promotes equitable distribution of computational power and data, reducing reliance on centralized entities and broadening AI accessibility.11 The tokenization of AI outputs transforms them into tangible, tradable assets within financial systems, enhancing liquidity for AI intellectual property and fostering innovative economic models, such as resource-sharing ventures and novel revenue streams.12

Smart contracts within blockchain enable self-executing transactions among AI systems, automating complex decision-making and interactions without necessitating human oversight. This automation ensures transactions adhere to predefined conditions, enhancing efficiency in domains like supply chain logistics and energy trading.13 As this author will demonstrate later in this article, AI agents are a significant extension of AI systems that can directly interface with smart contracts for transaction execution.

However, the juxtaposition of centralized and decentralized AI models presents a strategic conundrum. Centralized systems exploit vast data pools for efficiency, whereas decentralized frameworks promote innovation via data sovereignty and reduced monopolistic tendencies. Scholarly perspectives advocate a balanced approach to harness the strengths of both models.14

In practical contexts, the synergy of AI and blockchain demonstrates significant potential across supply chain management, financial systems, and cybersecurity. AI enhances process efficiency and data analytics, while blockchain ensures transparency and security.15

Despite these advancements, challenges such as scalability, resource intensity, and regulatory compliance persist, underscoring the need for continued research to fully realize this integration’s potential.16

In summation, the convergence of AI and blockchain transcends a simple technological merger, heralding a foundational shift towards a secure, transparent, and democratized digital infrastructure. This transformative potential demands ongoing interdisciplinary efforts to overcome current limitations and fully leverage its capacity to reshape industries and societal functions.17

# **3. AI Agents**

AI agents are autonomous software entities designed to execute specific tasks on behalf of their users, encompassing a convergence of computational engineering and cognitive simulation. AI agents represent a paradigm shift in autonomous computing, integrating sophisticated data processing, adaptive decision-making, and seamless system interoperability to serve user needs across diverse domains. Their ongoing evolution promises to further redefine the boundaries of artificial intelligence in practical applications.

AI agents leverage advanced algorithms, machine learning methodologies, and, in many instances, natural language processing capabilities to emulate human-like decision-making processes within digital ecosystems. Their deployment spans a broad spectrum of applications, from financial trading systems to personalized consumer services, reflecting their versatility and transformative potential in modern technology. The operational framework of AI agents starts with the interpretation of user directives, facilitated through diverse interfaces such as textual or vocal inputs, or through intricate analyses of behavioral patterns. This interpretive capacity is pivotal in ensuring that the actions undertaken by the agent align seamlessly with user expectations and objectives.18

Central to their functionality, AI agents aggregate and process data from an array of heterogeneous sources, encompassing real-time market feeds, detailed user profiles, and extensive historical transaction repositories. Machine learning models, often employing supervised and unsupervised techniques, are instrumental in extracting actionable insights from this voluminous data. This analytical prowess enables agents to identify patterns and correlations that inform subsequent decision-making.19

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW4 of 21

The decision-making apparatus of AI agents is underpinned by algorithms that oscillate between adherence to predefined rules and adaptive responses derived from learned patterns. In financial trading contexts, for instance, predictive models—such as regression analyses or neural networks—may be harnessed to optimize the timing of transactions or to calibrate pricing strategies, thereby maximizing economic outcomes. This adaptive capacity is critical in dynamic environments where market conditions fluctuate rapidly.20

To effectuate their decisions, AI agents interface with external digital systems through application programming interfaces (APIs), necessitating robust mechanisms for secure and authenticated communication. This integration ensures that transactions—whether financial, logistical, or informational—are executed with precision and integrity.21

Upon computing their decisions, AI agents operationalize these outcomes by executing transactions, ranging from trading securities in volatile markets to managing consumer purchases in e-commerce platforms. This execution phase underscores their role as intermediaries between abstract computation and tangible action.22 Following transaction execution, AI agents engage in continuous post-action monitoring, scrutinizing outcomes to refine their strategic approaches. This iterative process leverages feedback loops and newly acquired data to enhance the precision and efficacy of future decision-making. Such adaptability is a hallmark of advanced AI systems, distinguishing them from static rule-based programs.23 Finally, AI agents furnish users with comprehensive feedback, detailing the execution process, performance metrics, and any deviations from anticipated results. This transparency fosters trust and enables users to evaluate the agent’s efficacy critically.24

# _3.1. Applications_

AI agents are deployed across a variety of domains, exemplifying their adaptability and utility in automating complex tasks. In the realm of financial trading, AI agents leverage sophisticated market analysis to execute trades autonomously, capitalizing on real-time data and predictive modeling to optimize outcomes. These agents employ advanced techniques, such as high-frequency trading algorithms and microstructural analysis, to navigate volatile markets with precision.25

In the domain of e-commerce, personal shopping agents enhance the online purchasing experience by identifying optimal deals, curating product recommendations, and managing transactional logistics. These agents utilize recommender systems—often grounded in collaborative filtering or content-based approaches—to align offerings with user preferences, thereby streamlining decision-making and increasing satisfaction.26

Similarly, smart home agents exemplify the integration of AI into domestic environments, autonomously managing household transactions such as ordering supplies based on observed consumption patterns. These agents rely on sensor networks, machine learning, and predictive analytics to anticipate needs, thereby enhancing efficiency and convenience. For instance, they might reorder groceries or adjust energy usage in response to behavioral trends.27

# _3.2. Open Issues_

As these systems continue to evolve, balancing innovation with responsibility will remain paramount to harnessing their benefits while safeguarding societal values. The widespread adoption of AI agents necessitates stringent security protocols to protect sensitive user data and ensure the integrity of financial transactions. Robust encryption, secure API integrations, and authentication mechanisms are imperative to mitigate risks such as data breaches or unauthorized access. Beyond technical safeguards, the deployment of AI agents raises profound ethical considerations that demand careful scrutiny. Key among these are the imperatives of transparency—ensuring users

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW5 of 21

understand how decisions are made—securing informed consent for data usage, and addressing algorithmic bias to prevent inequitable outcomes. These issues are compounded by the potential for AI systems to inadvertently perpetuate societal disparities if trained on flawed datasets.28

By automating intricate decision-making and operational tasks, AI agents offer significant efficiencies and, in certain contexts, outperform human capabilities. Their ability to process vast datasets, adapt to dynamic conditions, and execute actions with precision positions them as invaluable tools in fields ranging from finance to domestic management. However, this enhanced performance must be tempered by rigorous design considerations to ensure that AI agents operate ethically and in alignment with user interests. Safeguards—such as regular audits of algorithmic behavior, mechanisms for user override, and transparent reporting—must be embedded within their architecture to prevent misuse or unintended consequences.

# **4. Cryptocurrency Payment Rails and AI Agents**

Cryptocurrency payment rails, rooted in blockchain technology, enable AI agents with a superior infrastructure for conducting financial transactions, outperforming the capabilities of traditional banking systems. This advantage derives, in part, from their decentralized design, programmable functionalities, and fortified security measures, which collectively augment the autonomy and efficiency of AI operations. Nevertheless, the confluence of AI and cryptocurrency payment rails presents significant monitoring challenges, necessitating advanced oversight mechanisms to ensure compliance, security, and ethical integrity.

Cryptocurrency payment rails function through decentralized blockchain networks, obviating the need for centralized intermediaries such as banks. This decentralization enables AI agents to autonomously oversee digital wallets—securing private keys, monitoring balances, and managing multiple addresses—without requiring human or institutional authorization.29 Conversely, traditional banking systems, reliant on centralized governance and manual validation, impose restrictions that curtail AI independence and engender operational delays.30 A pivotal attribute of these rails is the deployment of smart contracts—self-executing protocols embedded in the blockchain—which permit AI agents to automate transactions, such as releasing funds upon verified service completion through external data feeds (oracles).31 This programmability starkly contrasts with traditional banking’s dependence on less adaptable application programming interfaces (APIs), which lack the seamless integration inherent in blockchain frameworks.32

Moreover, cryptocurrency rails afford AI agents direct engagement with decentralized exchanges (DEXs), facilitating real-time trading of cryptocurrencies based on market conditions or preestablished algorithms.33 Traditional financial systems, hampered by broker intermediaries and limited market access, offer no analogous flexibility.34 The cryptographic underpinnings of blockchain bolster transaction security, empowering AI agents to safeguard user data effectively.35 Additionally, these agents can dynamically adjust transaction parameters in response to real-time market volatility or regulatory exigencies, such as adherence to Know Your Customer (KYC) and Anti-Money Laundering (AML) mandates.36 Traditional banking, burdened by centralized legacy infrastructure, exhibits diminished responsiveness to such exigencies.37

# **5. Monitoring Frameworks and Inherent Challenges**

The ecosystem supporting AI agents on cryptocurrency rails incorporates sophisticated monitoring tools. For example, platforms such as Coinbase utilize solutions like the CDP SDK Wallet Manager to ensure wallet security and transactional

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW6 of 21

compliance.38 Blockchain providers, including Biconomy, employ frameworks like the Delegated Authorization Network (DAN) to enforce permissioned AI activities.39 Compliance entities such as Chainalysis scrutinize transactions to identify illicit conduct, extending their purview to AI-driven operations.40 Developers and proprietors also establish internal mechanisms to align agent actions with legal and ethical standards.41

Despite these efforts, significant obstacles endure. The decentralized nature of blockchain complicates accountability, rendering oversight across distributed networks opaque.42 This opacity jeopardizes adequate regulatory supervision of AI agents.43 The accelerated evolution of AI and blockchain technologies frequently outstrips regulatory development, potentially situating agents within legal interstices, particularly in financial and data management domains.44 Furthermore, AI autonomy introduces unpredictability, wherein actions may diverge from intended outcomes, amplifying risks of unintended ramifications.45 Security vulnerabilities persist, as blockchain integration does not wholly preclude cyberattacks or privacy breaches absent rigorous monitoring.46

Prospectively, heightened regulatory scrutiny can be anticipated as AI-driven financial transactions proliferate, aiming to forestall fraud and ensure compliance.47 Specialized AI monitoring services and Decentralized Autonomous Organizations (DAOs) may arise, harnessing smart contracts for governance and risk mitigation.48 Self-monitoring among AI agents could enhance efficiency, though robust security protocols are imperative to thwart manipulation.49 Integration with expansive systems, such as the Internet of Things (IoT), may amplify monitoring capacities.50

# _5.1. Current Monitoring of AI Agents_

The integration of AI agents into cryptocurrency payment rails necessitates a complex monitoring framework to ensure the security, compliance, and operational integrity of autonomous financial transactions. To be clear, the existing monitoring framework is necessitated by the autonomous nature of AI agents, which leverage the decentralized and programmable infrastructure of blockchain-based cryptocurrency rails to conduct financial operations.51 The complexity of these systems, combined with their rapid evolution, underscores the critical role of coordinated oversight to mitigate risks such as illicit activity, regulatory noncompliance, and operational errors.52 The primary actors in this ecosystem—cryptocurrency exchanges, blockchain infrastructure providers, compliance and security services, and AI agent developers and owners—collectively form a robust yet intricate monitoring apparatus.53

For example, cryptocurrency exchanges, such as Coinbase, employ initiatives like the CDP SDK Wallet Manager to oversee AI-managed wallets, guaranteeing secure and compliant transactions.54 Blockchain infrastructure providers, exemplified by Biconomy, contribute through mechanisms like the Delegated Authorization Network (DAN), which enforces permissioned execution of AI transactions.55 Compliance and security services, notably Chainalysis, extend their analytical tools to monitor AI agent activities, focusing on the detection of illicit conduct and adherence to regulatory standards.56 Additionally, AI agent developers and owners implement internal systems to supervise their agents, ensuring alignment with intended functions and legal boundaries. Collectively, these actors form a multilayered monitoring ecosystem, critical to managing the complexities of AI autonomy in the rapidly evolving landscape of decentralized finance. This subchapter explores the roles, mechanisms, and implications of these oversight efforts, setting the stage for a deeper analysis of their efficacy and challenges.

# 5.1.1. Cryptocurrency Exchanges and Platforms

Cryptocurrency exchanges and platforms serve as pivotal entities in the monitoring of AI agent activities, leveraging their positions as gateways to blockchain networks to enforce security and compliance standards.57 For example, Coinbase, a leading

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW7 of 21

cryptocurrency exchange, has developed sophisticated tools such as the CDP SDK Wallet Manager, a software development kit designed to oversee AI-managed digital wallets.58 This initiative enables real-time tracking of wallet activities, ensuring that transactions executed by AI agents remain secure against unauthorized access and compliant with applicable regulatory frameworks.59 The CDP SDK Wallet Manager exemplifies how exchanges integrate advanced technological solutions to monitor AI operations, providing functionalities such as anomaly detection, balance verification, and adherence to Know Your Customer (KYC) protocols.60 By implementing such tools, exchanges not only safeguard their platforms but also contribute to the broader stability of the cryptocurrency ecosystem, addressing vulnerabilities inherent in decentralized financial systems.61

# 5.1.2. Blockchain Infrastructure Providers

Blockchain infrastructure providers constitute another critical layer in the monitoring framework, furnishing the technical scaffolding that enables secure and permissioned AI agent transactions.62 Biconomy, a prominent provider in this domain, employs its Delegated Authorization Network (DAN), which implements permissioned transaction controls, to ensure that AI agents operate within predefined parameters.63 The DAN functions as a decentralized authorization protocol, embedding permissions into the blockchain to regulate the scope of AI-driven transactions—such as specifying allowable counterparties, transaction limits, or execution conditions.64 This system ensures that AI agents adhere to their intended mandates, preventing unauthorized or erroneous activities that could compromise network integrity.65 By integrating such mechanisms, infrastructure providers like Biconomy enhance the reliability of cryptocurrency payment rails, bridging the gap between AI autonomy and operational accountability.66 Their role underscores the importance of foundational blockchain technologies in supporting scalable and secure AI applications.67

# 5.1.3. Compliance and Security Services

Compliance and security services further extend the monitoring landscape of AI agents by offering specialized tools to scrutinize cryptocurrency transactions, including those orchestrated by AI agents.68 Chainalysis, a leading blockchain analytics firm, provides analytical solutions that detect illicit activities—such as money laundering, fraud, or sanctions evasion—within blockchain networks.69 These tools extend to AI agent operations, enabling the identification of suspicious transactional patterns through advanced data analytics and blockchain forensics.70 Chainalysis’s focus on regulatory compliance ensures that AI-driven transactions align with legal standards, such as Anti-Money Laundering (AML) requirements, by cross-referencing transaction data against global watchlists and regulatory benchmarks.71 This external layer of oversight complements internal monitoring efforts, offering an independent check on AI behavior and reinforcing trust in decentralized financial systems.72 The integration of such services highlights the necessity of aligning technological innovation with regulatory imperatives.73

# 5.1.4. AI Agent Developers and Owners

Finally, AI agent developers and owners themselves play an indispensable role in the monitoring ecosystem, implementing internal systems to oversee their agents’ performance and ensure alignment with legal and ethical boundaries.74 These stakeholders deploy proprietary monitoring mechanisms—such as real-time activity logs, behavioral auditing tools, and fail-safe protocols—to verify that AI agents execute tasks as intended.75 For instance, developers may program agents with predefined operational limits, such as caps on transaction volumes or restrictions on interactions with high-risk entities, to mitigate potential misuse.76 Owners, meanwhile, maintain

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW8 of 21

continuous oversight to ensure compliance with jurisdictional laws and adherence to ethical guidelines, adapting agent functionalities in response to evolving standards.77 This internal accountability layer is critical, as it addresses the inherent unpredictability of AI autonomy, providing a first line of defense against operational deviations or unintended consequences.78 The proactive engagement of developers and owners thus forms a foundational component of the monitoring framework, complementing external efforts by exchanges, infrastructure providers, and compliance services.79

# _5.2. Expected Future Monitoring of AI Agents_

The prospective evolution of monitoring AI agents operating on cryptocurrency payment rails can be expected to undergo significant expansion and transformation, driven by technological advancements, regulatory imperatives, and systemic complexities.80

First, regulatory bodies can be expected to amplify their oversight as AI agents assume a more prominent role in financial transactions facilitated by cryptocurrency payment rails. The increasing prevalence of AI in executing autonomous trades, payments, and asset management amplifies the potential for fraud, market manipulation, and consumer harm, necessitating a fortified legal framework. This evolving regulatory environment, particularly surrounding cryptocurrencies, will demand stringent monitoring protocols to safeguard against illicit activities, ensure consumer protection through transparent and equitable transaction practices, and enforce compliance with financial regulations such as anti-money laundering (AML) and sanctions regimes.81 Such oversight may manifest in enhanced reporting requirements, real-time transaction audits, and mandatory integration of compliance tools into AI systems, reflecting a proactive governmental response to the proliferation of decentralized finance.82

Second, the emergence of specialized AI monitoring services represents a plausible advancement, concentrating exclusively on the supervision of AI-driven transactions.83 These bespoke services could harness advanced AI algorithms to predict risks—such as market volatility or agent malfunctions—detect anomalies in transactional behavior, and manage associated threats through automated interventions.84 By leveraging machine learning and predictive analytics, it is conceivable that these services might identify patterns indicative of fraud or non-compliance before they escalate, offering a preemptive rather than reactive approach to oversight.85 This development could spawn a novel market segment dedicated to AI monitoring tools, catering to cryptocurrency platforms, financial institutions, and regulatory bodies seeking scalable solutions to oversee autonomous agents.86

Third, Decentralized Autonomous Organizations (DAOs) are likely to evolve as governance and monitoring entities within blockchain ecosystems, exerting decentralized control over AI agents.87 DAOs, structured as community-governed entities operating via smart contracts, could encode rules into the blockchain to regulate AI behavior—such as restricting transaction types or enforcing ethical guidelines. Smart contracts might autonomously monitor agent activities by comparing them against predefined parameters, flagging deviations for review, and resolving disputes or instances of non-compliance through consensus-driven mechanisms. This decentralized approach could enhance transparency and accountability, reducing reliance on centralized authorities while aligning with the ethos of blockchain technology.

Fourth, AI self-monitoring emerges as a compelling future trajectory, wherein AI agents are designed to oversee one another’s operations within the cryptocurrency ecosystem.88 This paradigm envisions a network of interdependent agents, each equipped with monitoring protocols to audit peers’ transactions, validate compliance, and ensure operational fidelity.89 Such a system could enhance efficiency by distributing oversight tasks across agents, minimizing the need for external intervention.90 However,

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW9 of 21

this approach necessitates robust security measures—such as cryptographic safeguards and anti-collusion algorithms—to prevent coordinated manipulation or exploitation of vulnerabilities, ensuring that self-monitoring does not devolve into self-serving behavior.91

Fifth, the integration of AI agent monitoring with the Internet of Things (IoT) and broader interconnected systems could herald a more holistic approach to oversight.92 By linking cryptocurrency payment rails with IoT networks—such as smart devices, sensors, and data feeds—monitoring could extend beyond blockchain transactions to encompass real-world interactions, verifying the integrity of inputs, e.g., oracle data, and outputs, e.g., payment triggers.93 This convergence might enable a comprehensive ecosystem where connected devices collectively ensure transaction security, detect physical-world anomalies, and reinforce compliance across digital and tangible domains.94 Such an approach would leverage the scalability of IoT to address the expanding scope of AI applications in finance.95

The impetus for this anticipated increase in monitoring stems from multiple converging factors: heightened security concerns arising from AI vulnerabilities, the imperative of regulatory compliance amid evolving legal standards, the demand for consumer protection in decentralized markets, continuous innovations in AI and blockchain technology, and the scalability challenges of blockchain networks.96 As these domains progress, the methods and technologies for monitoring will adapt accordingly, incorporating new capabilities—such as real-time analytics and decentralized governance—while addressing emergent vulnerabilities like scalability bottlenecks or regulatory gaps.97 This dynamic evolution underscores the need for a forward-looking framework that balances innovation with oversight, ensuring the sustainable integration of AI agents into cryptocurrency ecosystems.98

# _5.3. AI-Based Agentic Monitoring_

The monitoring of AI agent-based transactions by centralized AI systems offers a multifaceted strategic approach to managing the security, compliance, and efficiency challenges inherent in modern digital infrastructures. By combining real-time monitoring, predictive modeling, regulatory adherence, advanced security protocols, comprehensive auditing, adaptive learning, user-friendly interfaces, and integrated system architectures, AI-based monitoring systems empower human owners with enhanced oversight capabilities.

# 5.3.1. Real-time Transaction Monitoring

AI systems can be designed to monitor AI agent transactions in real time by analyzing patterns and identifying anomalies within transaction data. Machine learning models, once trained on typical transactional behaviors, are capable of flagging deviations that may indicate errors or fraudulent activities. For instance, anomaly detection algorithms can examine variations in transaction frequency, volume, or the identities of counterparties to promptly alert system administrators to irregularities. This real-time capability is essential for maintaining operational continuity and safeguarding the integrity of digital transactions.99

# 5.3.2. Behavioral Analysis and Predictive Modeling

In addition to real-time Ai agent monitoring, AI systems can employ behavioral analysis to predict transaction outcomes based on historical data and observed agent behavior. By constructing predictive models that assess risk factors, these systems determine whether a transaction aligns with predetermined directives or poses potential threats. Techniques such as supervised learning and regression analysis enable AI systems to forecast success rates and flag non-compliant transactions before adverse outcomes occur. This proactive approach not only minimizes risk but also ensures that

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW10 of 21

transaction behaviors remain consistent with the owner’s strategic and regulatory objectives.100

# 5.3.3. Compliance and Regulatory Adherence

Ensuring that transactions adhere to established regulatory frameworks is a critical function of AI-based monitoring systems. These systems can be programmed to enforce compliance with standards such as Know Your Customer (KYC) and Anti-Money Laundering (AML) regulations. By incorporating Natural Language Processing (NLP) techniques, AI systems interpret complex legal texts and apply their provisions directly to transaction data. The immutable nature of blockchain technology further supports these efforts by providing a transparent, tamper-proof record of all transactions, which can be scrutinized for compliance.101

# 5.3.4. Security and Fraud Detection

Security remains a paramount concern in digital transactions, and AI systems contribute significantly to safeguarding operations through multi-layered security protocols. AI-driven intrusion detection systems learn from network patterns to distinguish between legitimate and malicious activities, thereby preemptively countering potential cyber threats. Additionally, simulation of potential attack vectors allows these systems to fortify defenses proactively. Such comprehensive security measures are critical for protecting the centralized repositories where AI agent transactions are processed and stored.102

# 5.3.5. Audit and Accountability

Post-transaction auditing is another critical function enabled by AI-based monitoring. Automated audit processes can cross-reference blockchain transaction records with internal logs to verify the integrity and consistency of transactions. Detailed reports generated by AI systems highlight discrepancies and potential areas of concern, ensuring transparency and accountability. Such audit trails are indispensable for regulatory compliance and for maintaining the trust of human owners overseeing AI agent activities.103

# 5.3.6. Adaptive Learning and System Improvement

AI-based monitoring systems are inherently capable of adaptive learning through the establishment of feedback loops that refine their operational parameters over time. By continuously analyzing past transactions, these systems adjust their detection algorithms to reduce false positives and improve the identification of emerging threats or inefficiencies. This dynamic recalibration is critical for maintaining high levels of accuracy and reliability in environments where transactional data volume and complexity are constantly evolving.104 However, the feedback loops in centralized AI learning systems as discussed herein need to be distinguished from web3-based AI feedback systems that are discussed later in this article.

# 5.3.7. User Interface and Interaction

For effective oversight, AI monitoring systems must present complex transactional data in a manner that is accessible and actionable for human owners. Advanced user interfaces, designed with the aid of AI-driven data interpretation techniques, translate intricate patterns and analytics into clear insights and alerts. This enhanced interaction facilitates intuitive decision-making and reduces the dependency on specialized technical knowledge, empowering owners to monitor and control AI agent transactions more effectively.105

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW11 of 21

## 5.3.8. Integration with Broader Systems

Finally, the efficacy of AI monitoring is amplified when integrated with broader Internet of Things (IoT) and enterprise systems. Such integration provides a holistic view of all transactions, enabling cross-system data correlation and comprehensive oversight of AI agent activities. This interconnected approach ensures that monitoring is not limited to isolated processes but extends to encompass the entire digital ecosystem, thereby maximizing security and operational efficiency.106

# **6. Shortcomings in AI Agent Monitoring**

## _6.1. Gaps in Current Monitoring Viability_

The existing framework for monitoring AI agents on cryptocurrency payment rails, while identifying key actors, falls short of delivering viable solutions. In particular, it fails to identify specifically scalability and adaptability challenges, and omits critical risks. These deficiencies across the ecosystem—cryptocurrency exchanges, blockchain infrastructure providers, compliance and security services, and AI agent developers and owners—reveal a structure unprepared for the complexities of decentralized finance and the ever evolving AI agents that scale DeFi systems.

The existing framework for monitoring AI agents using cryptocurrency payment rails offers no prescriptive measures—such as predictive analytics or cross-actor protocols—to anticipate evolving risks, rendering it reactive rather than proactive. Furthermore, it fails to articulate how the framework scales with the increasing volume and diversity of AI-driven transactions, leaving its operational resilience untested.

In particular, the identified drivers—security, compliance, consumer protection, innovation, and scalability—underscore the need for monitoring but are not matched with solutions adaptable to AI agent’s swift rise. The expected solutions for future AI monitoring fail to propose feedback-driven mechanisms,107 to balance innovation with oversight, leaving regulatory gaps and scalability bottlenecks unresolved. Without this anticipatory capacity, the framework cannot efficiently or fruitfully manage ubiquitous AI agents.

6.1.1. Scalability and Interoperability Deficits in Cryptocurrency Exchanges and Platforms

The discussion of exchange-based tools, such as the CDP SDK Wallet Manager, lacks detail on scalability under high transaction loads or adaptability across heterogeneous blockchain protocols. It does not address how these tools counter sophisticated threats, like adversarial AI agents exploiting wallet vulnerabilities, nor evaluate their feasibility for smaller exchanges, limiting broader applicability. The absence of interoperability considerations with non-standardized networks further restricts its utility in a fragmented ever evolving DeFi landscape.

## 6.1.2. Adaptability and Risk Mitigation in Blockchain Infrastructure Providers

The portrayal of infrastructure mechanisms, like the DAN, omits specifics on adapting to diverse AI agent designs or scaling across complex blockchain networks. It also neglects critical risks, such as smart contract exploits, bugs, or permission conflicts, and fails to propose real-time enforcement solutions across distributed nodes.108 This lack of granularity and risk mitigation weakens its claim to bridge AI autonomy and accountability.

## 6.1.3. Scope and Real-Time Limitations in Compliance and Security Services

The reliance on compliance tools, such as those from Chainalysis, overlooks their scalability challenges in monitoring vast, decentralized transaction volumes and adaptability to jurisdictional regulatory disparities.109 It fails to address latency in

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW12 of 21

forensic analysis, limiting real-time detection, and provides no strategy for overseeing transactions on privacy-focused blockchains where opacity hampers traditional methods.110 This narrow focus undermines its effectiveness as a comprehensive oversight solution.111

# 6.1.4. No Standardization and Accountability in AI Agent Developers and Owners

The emphasis on internal monitoring by developers and owners lacks a framework for standardizing proprietary systems across diverse stakeholders, risking inconsistent oversight.112 It does not address accountability mechanisms for agent deviations, such as insider manipulation or third-party agent risks, nor propose auditing standards against external benchmarks.113 This omission leaves internal efforts fragmented and unreliable.114

# _6.2. Shortcomings of Expected Future Monitoring Solutions_

While the prospective framework for monitoring AI agents on cryptocurrency payment rails anticipates transformative developments driven by technological, regulatory, and systemic dynamics, it falls critically short of offering viable solutions. Its inability to suggest a viable solution can be traced predominantly back to the rapid evolution and impending ubiquity of AI agents.

As AI agents proliferate and adapt at an unprecedented pace, outstripping static oversight mechanisms, the prospective framework for monitoring AI agents on cryptocurrency payment rails speculative projections lack the specificity, adaptability, and anticipatory capacity required for effective monitoring.115

# 6.2.1. Regulatory Oversight: Outpaced by AI Evolution

The expectation of enhanced regulatory oversight fails to account for the accelerating evolution of AI agents, which will soon dominate financial transactions, rendering static legal frameworks obsolete. The proposed framework’s vague references to audits and compliance tools lack mechanisms to adapt to AI’s adaptive capabilities, such as self-optimizing algorithms that evade traditional detection. Anticipatory regulation requires feedback loops—continuous data inputs from AI behavior—to dynamically update rules, a necessity unmet here given the absence of such adaptive systems.116 Without feedback loops, regulatory efforts cannot efficiently address fraud or consumer harm as AI ubiquity amplifies these risks across decentralized networks.

# 6.2.2. Specialized AI Monitoring Services: Static Tools in a Dynamic Landscape

The proposed specialized AI monitoring services, while conceptually promising, do not grapple with the rapid adaptability of ubiquitous AI agents, which demand real-time, evolving oversight. It is entirely unclear how these specialized AI monitoring services will scale computationally or adjust algorithms as AI agents diversify, a critical flaw given their anticipated pervasiveness. This static vision lacks the efficiency needed for preempting risks.

# 6.2.3. Traditional DAOs - Overwhelmed by AI Agent Ubiquity

The evolution of DAOs as monitoring entities assumes a static governance model117 incapable of keeping pace with the rapid proliferation and sophistication of AI agents. It remains unclear how DAOs will manage the computational burden of overseeing ubiquitous AI agents or adapt smart contracts to their evolving behaviors. Feedback loops are essential for DAOs to dynamically adjust their governance rules based on AI activity, a mechanism absent here, risking inefficiency and vulnerability to exploits. Without this adaptability, DAOs cannot fruitfully govern an AI-dominated ecosystem.

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW13 of 21

# 6.2.4. AI Self-Monitoring: Unchecked Evolution Undermines Security

AI self-monitoring is envisioned as efficient but fails to address how rapidly evolving agents will maintain integrity as they become ubiquitous. The proposed reliance on unspecified security measures overlooks the risk of adaptive AI agents colluding or evading oversight, a concern amplified by their pervasive deployment. Feedback loops are crucial to monitor AI agent interactions in real time, ensuring accountability—an approach absent here, rendering the system susceptible to manipulation. This lack of dynamic safeguards jeopardizes its viability.

# 6.2.5. IoT Integration: Scalability Outstripped by AI Pace

The integration with IoT aims for holistic oversight but does not account for the pace at which AI agents will outgrow static IoT-blockchain linkages. The text omits how this convergence will handle latency or secure data as AI ubiquity drives exponential transaction complexity. The governance feedback loop model—integrating real-time IoT data to adapt monitoring118 —is essential yet unaddressed, leaving the approach inefficient against AI’s rapid evolution. This gap undermines its practicality.

# _6.3. Fallacy of Centralized AI Monitoring AI Agent Evolution_

The abovementioned centralized AI-driven monitoring of AI agents119 encounters fallacies that undermine its capacity to effectively oversee the rapid evolution of AI agents.120 Centralized AI structures for AI agent monitoring create a self-referential loop prone to systemic biases and blind spots, while their rigidity fails to adapt to AI’s dynamic nature.

# 6.3.1. Circular Dependency: AI Monitoring AI Creates Inherent Bias

The foundational premise of AI systems monitoring AI agent transactions introduces a fallacy, as the monitoring AI inherits the same adaptive traits and potential flaws as the agents it oversees.121 Real-time anomaly detection and behavioral analysis rely on machine learning models trained on agent data, which may replicate biases or fail to detect novel deviations not present in training sets.122 This self-referential loop lacks an external benchmark to ensure objectivity, which in turn undermines its reliability as AI agents evolve beyond initial parameters.123

# 6.3.2. Insufficient Adaptability: Static Centralization vs. Dynamic Evolution

Centralized AI monitoring systems, despite adaptive learning claims, cannot sufficiently keep pace with AI agents’ rapid evolution due to their hierarchical, top-down design.124 Centralized AI monitoring schemes’ reliance on feedback loops within a centralized framework adjusts algorithms retrospectively, but lacks the anticipatory capacity. Real-time, decentralized data inputs—to match AI’s exponential adaptability are essential.125 As AI agents proliferate across decentralized networks, static centralized models fail to scale or adapt to diverse, emergent behaviors.126

Moreover, the assertion that centralized AI ensures compliance with KYC and AML standards overlooks the circularity of relying on AI to interpret regulations it may itself violate.127 NLP-driven compliance assumes static legal frameworks, yet AI agents’ evolution introduces novel transaction types that outstrip predefined rules, undetected by centralized systems lacking external validation. Feedback loops, requiring diverse stakeholder inputs, are essential for AI agent monitoring and highlight the insufficiency of centralized AI oversight to dynamically align with evolving standards.

# 6.3.3. Security and Fraud Detection: Centralized Vulnerabilities Amplify Risks

Centralized AI-driven security protocols, while robust against known threats, are insufficient because of their single-point-of-failure design, which AI agents can exploit as

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW14 of 21

they evolve.128 The circular reliance on AI to simulate attack vectors assumes it can anticipate its own adaptive strategies—an inherent fallacy as evolving agents may bypass centralized defenses.129

The audit function’s dependence on AI to cross-reference blockchain records creates circular verification, insufficient for ensuring accountability as AI agents evolve.130 Centralized systems lack external checks to validate AI-generated reports, risking undetected errors or manipulations by adaptive agents.131 Decentralized feedback loops, incorporating diverse perspectives,132 underscores the need for independent oversight missing in this approach.

## 6.3.4. Adaptive Learning: Centralized Constraints Limit Evolution

The centralized AI adaptive learning touted as a strength is insufficient, as it relies on internal data loops that cannot match AI agents’ external evolution.133 This circular recalibration fails to incorporate real-time, ecosystem-wide inputs that are essential for anticipating emerging threats. As AI agents become ubiquitous, centralized AI monitoring systems lag behind, unable to scale dynamically lock step with AI agent evolution.134

Enhanced user interfaces and IoT integration, while user-friendly, exacerbate centralized AI monitoring of AI agents because interfaces create data bottlenecks as AI agents proliferate.135 The circular reliance on AI to interpret complex data assumes static scalability, ignoring the exponential growth of transactions.136 Decentralized systems can handle these issues better through distributed processing as discussed further below.

# **7. Decentralized AI Agent Monitoring Solutions**

The proposed decentralized governance system for monitoring AI agent transaction execution—achieved through the integration of web3 community software with federated communication platforms—presents a robust alternative to traditional AI-driven supervision. The proposed system for monitoring AI agent transaction execution—integrating web3 community software, federated communication platforms, blockchain, smart contracts, WDAGs, and community-driven validation mechanisms—offers significant advantages over conventional, centralized AI-driven supervision.

By ensuring transparency and accountability, facilitating decentralized decision-making, and promoting real-time, automated responses, the proposed system not only mitigates inherent risks associated with centralization but also fosters an inclusive, and robust governance framework. This holistic approach aligns AI operations with both regulatory frameworks and community values, setting new standards for the development and deployment of AI technologies. Collectively, these advantages demonstrate that a decentralized governance model is superior to AI-driven supervision for ensuring the secure, compliant, and efficient execution of AI agent transactions in modern digital infrastructures.

## _7.1. Transparency and Accountability_

At the heart of the proposed decentralized AI agent monitoring system lies blockchain technology, which guarantees that every decision and transaction is recorded on an immutable ledger. This permanent, tamper-proof record enables public verification and fosters accountability by ensuring that no single entity can alter the historical record of AI agent activities. In contrast, AI-driven supervision, often operating within centralized frameworks, can obscure accountability through opaque decision-making processes and data handling. The blockchain-based model, by contrast, enhances trust among stakeholders because every action is transparently logged and subject to independent audit.137

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW15 of 21

# _7.2. Decentralized Decision-Making_

Unlike centralized AI agent monitoring—which relies on a limited number of systems or algorithms making decisions—the proposed decentralized AI agent monitoring model leverages the distributed nature of web3 technology to facilitate consensus-driven decision-making among a diverse group of stakeholders. This decentralized approach minimizes the risk of bias and single points of failure, as governance decisions are made collectively rather than being subject to the limitations or potential errors of a solitary AI system. This inclusive mechanism improves the overall integrity of transaction monitoring by incorporating varied perspectives and expertise.138

# _7.3. Web3 Governance via Federated Communication for Operational Efficiency_

The proposed integration of federated communication platforms such as Matrix with its Synapse server distributes data processing tasks across multiple nodes. This federated model not only scales the monitoring of AI transactions efficiently but also maintains data privacy by processing sensitive information locally. In contrast, centralized AI monitoring systems typically require aggregating data into a single repository, which creates a vulnerability by exposing a single point of failure and potentially compromising sensitive data. The federated approach thus offers improved operational efficiency and enhanced privacy protections.139

# _7.4. Smart Contracts for Automated Governance_

A significant advantage of the proposed web3 governance system is the utilization of smart contracts, which automate compliance checks, performance validations, and other governance tasks by executing predefined rules. These contracts execute automatically when pre-defined self enforcing conditions are met, ensuring that AI agent transactions are continuously monitored and managed according to community-established rules. The automated nature of smart contracts minimizes the need for manual oversight, reduces the potential for human error, and guarantees a consistent application of governance protocols—features that are often lacking in purely AI-driven supervisory models, though consensus delays may occur.140

# _7.5. Validation Pools and Reputation Tokens_

The proposed system of web3 governance for AI agent monitoring further enhances decentralized oversight through the implementation of Validation Pools and Reputation (REP) tokens. Validation Pools are mechanisms in which members stake their reputation tokens to vote on the approval or disapproval of transactions, proposals, or activities. REP tokens are non-transferable tokens that represent a user's trustworthiness and influence. They affect governance participation, profit sharing, and overall decision-making power. Within these pools, community members stake REP tokens to evaluate the quality and compliance of AI agent transactions. The resulting consensus not only influences the minting of REP tokens but also serves as a decentralized method of quality control. This mechanism ensures that monitoring is driven by collective expertise and community considerations rather than solely by algorithmic determinations, thereby reducing systemic bias and increasing the legitimacy of governance decisions.141

# _7.6. WDAGs for Structured Governance_

WDAGs provide a structured framework for mapping dependencies and establishing precedence among governance elements. By representing each governance rule or decision as a node and their relationships as weighted directed edges, WDAGs create a dynamic, traceable, and adaptable governance structure.142 The acyclic nature ensures no circular dependencies exist, preventing governance deadlocks, while weights

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW16 of 21

define the relative importance of connections between nodes. This approach allows for the efficient updating of protocols in response to new ethical or legal standards—an adaptability that is typically more limited in AI-driven supervision systems, which may operate on static algorithmic rules.143

# _7.7. Decentralized Community Engagement_

Central to the proposed system is the active participation of a diverse community in the governance process. By enabling community-driven decision-making, the system incorporates a wide range of perspectives, which are essential for monitoring the societal impacts of AI agent transactions. This participatory approach contrasts sharply with centralized AI-driven monitoring systems, which may lack the necessary diversity in oversight and can inadvertently perpetuate narrow, biased perspectives.144 The combination of smart contracts and WDAGs in the proposed system enables real-time monitoring and rapid response to anomalies in AI agent transactions. This dynamic responsiveness is crucial for maintaining compliance and adapting to emerging threats or regulatory changes. While AI-driven supervision systems can detect anomalies, their centralized nature may delay response times or obscure the decision-making process.145 The proposed decentralized framework, by contrast, ensures immediate corrective action through automated and transparent mechanisms.

# _7.8. Regulatory Oversight: Adapting to AI Evolution with Feedback Loops_

The critique that regulatory oversight cannot keep pace with AI agents’ accelerating evolution, rendering static frameworks obsolete, is addressed by the proposed DAO-centric dynamic web3 governance model. The WDAG structure, with its on-chain forum and validation pools, enables continuous updates to governance rules through expert community consensus, incorporating real-time AI behavior data. Meanwhile, feedback loops—integrating live inputs to adjust regulations dynamically—ensuring adaptability to self-optimizing AI algorithms. By automating the integration of evolving standards via smart contracts, the system efficiently mitigates fraud and consumer harm across ubiquitous AI deployments.

# _7.9. Dynamic Scalability Through Decentralized Tools_

The limitation of specialized AI monitoring services as static tools unable to adapt to AI agents’ rapid diversification is overcome by the DAO’s decentralized, scalable architecture. The WDAG-based forum and validation pools leverage community-driven analytics, scaling computationally via roll-ups—off-chain activities consolidated on-chain—to handle pervasive AI transactions. The proposed feedback loop model supports this by enabling real-time risk prediction and adjustment, ensuring efficiency as AI agents evolve. This dynamic system preempts risks, surpassing the static vision critiqued.

# _7.10. Enhanced Governance via WDAG Adaptability_

The concern that traditional DAOs are overwhelmed by AI ubiquity due to static governance is rectified by the proposed system’s WDAG-enhanced DAO. The WDAG’s acyclic, weighted structure dynamically integrates new AI behaviors as posts, adapting smart contracts through validation pools to manage computational loads. The proposed model’s emphasis on feedback loops ensures DAOs evolve rules based on real-time AI activity, reducing vulnerability to exploits and enabling fruitful governance of an AI-dominated ecosystem. This adaptability addresses the scalability and flexibility gaps identified.

# _7.11. Securing Integrity with WDAG Oversight_

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW17 of 21

The risk of unchecked AI self-monitoring, where evolving agents might collude or evade oversight, is mitigated by the DAO’s WDAG framework. By treating AI models as posts within the WDAG, linked to governance precedents via weighted edges, the system enforces accountability through community-validated smart contracts. The proposed feedback loops monitor interactions in real time, preventing manipulation with cryptographic safeguards embedded in the blockchain. This dynamic oversight ensures viability as AI agents become ubiquitous.

## _7.12. Scalable Convergence via Real-Time Data_

The shortfall in IoT integration—unable to scale with AI agents’ rapid pace—is resolved by the DAO’s integration of federated communications and WDAG analytics. The Matrix platform and real-time IoT data feeds, processed through the WDAG, address latency and secure data across complex transactions. The proposed governance feedback loops adapt monitoring using live inputs, ensuring efficiency against AI’s exponential growth. This practical, scalable solution overcomes the static linkage critique.

The proposed dynamic evolutionary DAO-centric web 3 governance system, with its WDAG-based governance, directly addresses the monitoring issues by offering a dynamic, scalable, and anticipatory framework. Leveraging feedback loops—as author has advocated for almost a decade—it adapts to AI agents’ rapid evolution and ubiquity, ensuring efficient, fruitful oversight.

# **8. Conclusion**

In conclusion, this paper has shown the inherent limitations of traditional, centralized AI-driven monitoring systems in overseeing the transaction execution of AI agents, particularly within the rapidly evolving landscape of decentralized digital infrastructures. These conventional approaches, characterized by opacity, susceptibility to bias, and single points of failure, are increasingly inadequate in addressing the complexities introduced by AI agents operating on cryptocurrency payment rails. In response, the proposed decentralized governance model, rooted in a web3 DAO-centric framework, emerges as a transformative solution, offering a robust, transparent, and adaptive alternative that aligns technological oversight with both community values and regulatory imperatives.

Looking forward, the web3 DAO-centric governance system provides innovative solutions to the multifaceted challenges of AI agent monitoring. By harnessing blockchain’s immutable ledger, smart contracts, and WDAGs, the proposed model ensures real-time transparency, automated compliance, and structured adaptability, effectively mitigating risks such as fraud, regulatory lag, and systemic bias. The integration of federated communication platforms and validation pools further enhances scalability and privacy, distributing oversight across a diverse stakeholder community and fostering a resilient ecosystem capable of evolving alongside AI advancements. Crucially, the incorporation of feedback loops—drawing on real-time data and community consensus—enables the system to anticipate and adapt to the accelerating pace of AI agent proliferation, addressing scalability bottlenecks and interoperability deficits that plague current frameworks.

This forward-looking approach not only overcomes the shortcomings of centralized and static monitoring solutions but also sets a new standard for the future of AI governance. By empowering decentralized communities to actively shape oversight protocols, the proposed system ensures that AI agent transactions remain secure, compliant, and efficient, even as their ubiquity transforms digital and economic landscapes. As AI and blockchain technologies continue to converge, the web3 DAO-centric model offers a scalable, democratic, and anticipatory framework that

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW18 of 21

promises to safeguard societal interests while unlocking the full potential of autonomous digital systems. Future research and implementation efforts should focus on refining these mechanisms, ensuring their practical deployment, and fostering interdisciplinary collaboration to realize a governance paradigm that thrives in an AI-driven world.

# **References**

1. Professor of Law. The author is grateful for excellent research assistance from Michael Bernardi, research assistant.

2. _See, e.g._ , Haoxiang Luo et al., _A Perspective of Trusted Artificial Intelligence When Blockchain Meets Large Language Models_ , 599 NEUROCOMPUTING 1 (2024), https://doi.org/10.1016/j.neucom.2024.128089.

3. See, e.g., Zhonghua Zhang et al., Recent Advances in Blockchain and Artificial Intelligence Integration: Feasibility Analysis, Research Issues, Applications, Challenges, and Future Work, 2021 SEC. & COMM. NETWORKS 1, 4-5 (2021), https://doi.org/10.1155/2021/9991535.

4. _See, e.g._ , Sandner et al., _Convergence of Blockchain, IoT, and AI_ , 3 FRONTIERS BLOCKCHAIN, at 3-5 (2020), https://doi.org/10.3389/fbloc.2020.522600.

5. _See, e.g._ , Nitine Rane et al., _Blockchain and Artificial Intelligence (AI) Integration for Revolutionizing Security and Transparency in Finance_ 1, 10-11 (SSRN, Dec. 2023), http://dx.doi.org/10.2139/ssrn.4644253.

6. _See, e.g._ , Nakhoon Choi & Heeyoul Kim _, Technological Convergence of Blockchain and Artificial Intelligence: A Review and Challenges_ , 14 ELECTRONICS 84, 23-24 (2025), https://doi.org/10.3390/electronics14010084.

7. _See, e.g._ , Satish Kumar et al., _Artificial Intelligence and Blockchain Integration in Business: Trends from a Bibliometric-Content Analysis_ , 25 INFO. SYS. FRONTIER 871, 892-893 (2023), https://doi.org/10.1007/s10796-022-10279-0.

8. _See, e.g._ , Luo et al. _, supra_ note 2 at, 4, 8, & 10.

9. _See, e.g._ , Abdullah Ayub Khan et al _._ , _The Collaborative Role of Blockchain, Artificial Intelligence, and Industrial Internet of Things in Digitalization of Small and Medium-size Enterprises_ , 13 SCIENTIFIC REPORTS 1656, 1-2 (2023), https://doi.org/10.1038/s41598-023-28707-9.

10. _See, e.g._ , Mark C. Ballandies et al., _A Taxonomy for Blockchain-based Decentralized Physical Infrastructure Networks (DePIN)_ , 2023 IEEE WORLD FORUM INTERNET THINGS at 1 (2023), https://doi.org/10.1109/WF-IoT58464.2023.10539514.

11. _See, e.g._ , _Zhibin Lin et al., Decentralized Physical Infrastructure Network (DePIN): Challenges and Opportunities_ , ARXIV 2406.02239v1 at 2-4, 6-7 (June 2024), https://arxiv.org/html/2406.02239v1.

12. _See, e.g._ , Alejandro Baldominos & Yago Saez _, Coin.AI: A Proof-of-Useful-Work Scheme for Blockchain-Based Distributed Deep Learning_ , 8 ENTROPY 723, at 7-14 (2019), https://doi.org/10.3390/e21080723.

13. _See, e.g._ , Mohsen Soori et al., _AI-powered Blockchain Technology in Industry 4.0, a Review_ , 1 J. ECONOMY & TECH. 222, 223-226, 231-232 (2023), https://doi.org/10.1016/j.ject.2024.01.001.

14. _See, e.g._ , Xiang Hui & Catherine Tucker, _Decentralization, Blockchain, Artificial Intelligence (AI): Challenges and Opportunities_ , at 17-22 (SSRN, Dec. 2024), http://dx.doi.org/10.2139/ssrn.4601788.

15. _See, e.g._ , Vincent Charles et al., _A Critical Analysis of the Integration of Blockchain and Artificial Intelligence for Supply Chain_ , 327 ANNALS OPERATION RSCH. 7, at 3-9 (2023), https://doi.org/10.1007/s10479-023-05169-w; Nitine Rane et al., _supra_ note 5.

16. _See, e.g._ , Moetez Abdelhamid et al., _A Review on Blockchain Technology, Current Challenges, and AI-Driven Solutions_ , 57 ACM COMPUTING SURVEYS 1 (2024), https://doi.org/10.1145/3700641. Moreover, regulatory frameworks must evolve concurrently with technological progress to prevent misuse and uphold ethical standards. _See, e.g._ , Oleksandr Kuznetsov et al., _On the Integration of Artificial Intelligence and Blockchain Technology: A Perspective About Security_ , 12 IEEE ACCESS 3881 (2024), https://doi.org/10.1109/ACCESS.2023.3349019; Ahmed M. Shamsan Saleh _, Blockchain for Secure and Decentralized Artificial Intelligence in Cybersecurity: A Comprehensive Review_ , 5 BLOCKCHAIN: RSCH. & APPLICATIONS (2024), https://doi.org/10.1016/j.bcra.2024.100193.

17. _See, e.g._ , Al Montha Al Shareef et al., _Integration of Blockchain with Artificial Intelligence Technologies in the Energy Sector: A Systematic Review_ , 12 FRONTIERS IN ENERGY RSCH., at 1-7 (2024), https://doi.org/10.3389/fenrg.2024.1377950.

18. For a foundational exploration of these principles, _see_ STUART J. RUSSELL & PETER NORVIG, ARTIFICIAL INTELLIGENCE: A MODERN APPROACH, at 31–34 (3d ed. 2010) (detailing the theoretical underpinnings of intelligent agents and their interaction with users).

19. For an in-depth examination of these techniques, _see_ IAN H. WITTEN ET AL., DATA MINING: PRACTICAL MACHINE LEARNING TOOLS AND TECHNIQUES 353–376 (3rd ed. 2016) (elucidating the application of machine learning in data-intensive environments).

20. _See_ EFRAIM TURBAN ET AL., ELECTRONIC COMMERCE 2018: A MANAGERIAL AND SOCIAL NETWORKS PERSPECTIVE 191–195 (9th ed. 2018) (discussing predictive analytics in e-commerce and trading systems).

21. The technological infrastructure supporting such interactions is rigorously examined in Stefan Tilkov & Steve Vinoski, _Node.js: Using JavaScript to Build High-Performance Network Programs_ , 14 IEEE INTERNET COMPUTING 80, 81-83 (2010) (exploring scalable network programming paradigms) https://www.doi.org/10.1109/MIC.2010.145.

22. _See e.g._ , Fatima Dakalbab et al., _Artificial intelligence techniques in financial trading: A systematic literature review_ , 36 J. KING SAUD UNIV. – COMPUTER INFO. SCIENCES (Article No. 102015), at 3-6, https://doi.org/10.1016/j.jksuci.2024.102015 (for an analysis of AI in trading); _see also_ , _AI agents for retail and e-commerce: Capabilities, components, use cases, implementation, and benefits_ , LEEWAYHERTZ,

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW19 of 21

https://www.leewayhertz.com/ai-agents-in-retail-and-ecommerce/ (last visited, Mar. 2, 2025) (for an analysis of AI agents in retail and e-commerce)

23. _See_ Andrew Y. Ng & Michael I. Jordan, _On Discriminative vs. Generative Classifiers: A Comparison of Logistic Regression and Naive Bayes_ , ADVANCES NEURAL INFO. PROCESSING SYSTEMS 841–848 (2002) (comparing adaptive learning models) https://papers.nips.cc/paper_files/paper/2001/hash/7b7a53e239400a13bd6be6c91c4f6c4e-Abstract.html.

24. For an applied perspective on feedback mechanisms in data-driven systems, _see_ DURSUN DELEN, REAL-WORLD DATA MINING: APPLIED BUSINESS ANALYTICS AND DECISION MAKING 126–129 (FT Press 2014) (highlighting the role of analytics in performance reporting).

25. For an analysis of market microstructure principles, including performance benchmarking and timing strategies relevant to high-frequency algorithmic trading, _see_ LARRY HARRIS, TRADING AND EXCHANGES: MARKET MICROSTRUCTURE FOR PRACTITIONERS 451–460 (Oxford Univ. Press 2003) (elucidating the role of automated systems in modern financial markets).

26. This application is explored in J. Ben Schafer et al., _Recommender Systems in E-Commerce_ , PROC. 1ST ACM CONF. ELEC. COM. 158 (1999) (detailing the algorithmic foundations of personalized shopping agents) https://doi.org/10.1145/336992.337035.

27. This functionality is analyzed in DIANE J. COOK & SAJAL K. DAS, SMART ENVIRONMENTS: TECHNOLOGY, PROTOCOLS, AND APPLICATIONS 135–148 (John Wiley & Sons 2004) (discussing the technological infrastructure of intelligent home systems).

28. For a philosophical and practical exploration of these challenges, _see_ LUCIANO FLORIDI, THE ETHICS OF INFORMATION 76–82 (Oxford Univ. Press 2013) (examining the ethical dimensions of informational technologies, including AI and arguing that ethical frameworks must evolve alongside technological advancements to ensure accountability and fairness, a perspective increasingly relevant as AI agents assume greater autonomy).

29. Vitalik Buterin _, Ethereum: A Next-Generation Smart Contract and Decentralized Application Platform_ , ETHEREUM 5-6 (2013), https://ethereum.org/en/whitepaper/.

30. ARVIND NARAYANAN ET AL., BITCOIN AND CRYPTOCURRENCY TECHNOLOGIES: A COMPREHENSIVE INTRODUCTION 123-126 (Princeton University Press 2016).

31. CRAIG CALCATERRA & WULF A. KAAL, DECENTRALIZATION: TECHNOLOGY’S IMPACT ON ORGANIZATIONAL AND SOCIETAL STRUCTURE 54-55 (2021).

32. NARAYANAN ET AL., _supra_ note 30, at 145-148.

33. Fabian Schär _, Decentralized Finance: On Blockchain- and Smart Contract-Based Financial Markets_ FED. RSRV. BANK ST. LOUIS REV. 153, 157-159 (2021), https://doi.org/10.20955/r.103.153-74.

34. PRIMAVERA DE FILIPPI & AARON WRIGHT, BLOCKCHAIN AND THE LAW: THE RULE OF CODE 78-80 (Har. University Press 2018).

35. Eli Ben-Sasson et al., _Zerocash: Decentralized Anonymous Payments from Bitcoin_ , 2014 IEEE SYMP. SEC. PRIV. 459, 461-462 (2014), https://doi.org/10.1109/SP.2014.36.

36. _Id._ at 462-463.

37. NARAYANAN ET AL., _supra_ note 30, at 130-132.

38. Justin, _AI Agents on Base_ BLOCKBASE (Oct. 25, 2024), https://insights.blockbase.co/ai-agents-on-base/.

39. Ana Paula Pereira, _Biconomy Onboards AI Agents for Onchain Transactions_ , COINTELEGRAPH (Jun. 11, 2024), https://cointelegraph.com/news/biconomy-ai-agents-on-chain-transactions.

40. CHAINALYSIS, https://www.chainalysis.com/ (last visited Feb. 27, 2025).

41. MASSIMO BARTOLETTI ET AL., _An Empirical Analysis of Smart Contracts: Platforms, Applications, and Design Patterns_ , in FIN. CRYPTOGRAPHY AND DATA SEC. 494, 502-504 (Michael Brenner et al. eds., 2017) (Lecture Notes in Comput. Sci. Vol. 10323).

42. Ballandies et al., _supra_ note 10, at 5.

43. Zhibin Lin et al., _Decentralized Physical Infrastructure Networks (DePIN): Challenges and Opportunities_ , IEEE NETWORK, at 6 (Early Access Oct. 29, 2024), https://doi.org/10.1109/MNET.2024.3487924.

44. DE FILIPPI & WRIGHT, _supra_ note 34, at 150-152.

45. NICK BOSTROM, SUPERINTELLIGENCE: PATHS, DANGERS, STRATEGIES 115-118, 138-140 (Oxford Univ. Press 2014).

46. Kuznetsov et al., _supra_ note 16, at 3881-3882.

47. DE FILIPPI & WRIGHT, _supra_ note 34, at 165-167.

48. Calcaterra & Kaal, _supra_ note 31, at 74; _see also_ Aaron Wright & Primavera De Filippi _, Decentralized Blockchain Technology and the Rise of Lex Cryptographia_ , at 16 (Mar. 12, 2015) (unpublished manuscript), http://dx.doi.org/10.2139/ssrn.2580664.

49. BOSTROM, _supra_ note 45, at 137-138, 140-142.

50. Kuznetsov et al., _supra_ note 16, at 3885-3886.

51. ANDREAS M. ANTONOPOULOS, MASTERING BITCOIN: PROGRAMMING THE OPEN BLOCKCHAIN 236-237 (1d ed. 2014).

52. KEVIN WERBACH, THE BLOCKCHAIN AND THE NEW ARCHITECTURE OF TRUST 132-135 (MIT Press 2018).

53. Joshua de Vos & Jacob Joseph, _CoinDesk 20 Quarterly Report Q4 2024 Market Commentary and Analysis_ , COINDESK 18-20 (2024), https://downloads.coindesk.com/cd3/CDI/CoinDesk+20+Quarterly+Report.pdf.

54. Justin, _supra_ note 38.

55. Pereira, _supra_ note 39.

56. CHAINALYSIS, _supra_ note 40.

57. Nathan Reiff _, Cryptocurrency Exchanges: What They Are and How to Choose, Investopedia_ (Nov. 22, 2024), https://www.investopedia.com/tech/190-cryptocurrency-exchanges-so-how-choose/.

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW20 of 21

58. COINBASE, _Coinbase Developer Platform: Wallet Command Center_ , https://docs.cdp.coinbase.com/learn/docs/wallet-command-center (last visited Feb. 28, 2025).

_59. Id. 60. Id._ 61. Chainalysis Team, _Introduction to Cryptocurrency Exchange Compliance_ , CHAINALYSIS (Nov. 21, 2024), https://www.chainalysis.com/blog/introduction-to-cryptocurrency-exchange-compliance-crypto-businesses-2024/.

62. Buterin, _supra_ note 29, at 13. 63. _Delegated Authorization Network,_ BICONOMY, https://www.biconomy.io/delegated-authorisation-network (last visited Feb. 28, 2025).

64. _Id. 65. Id._ 66. _Securing the AI-Driven Crypto Future: How Biconomy's Permissions Infrastructure Empowers Users_ , BICONOMY (Dec. 16, 2024), https://www.biconomy.io/post/securing-the-artificial-intelligence-driven-crypto-future-biconomy-permissions-infrastructure.

_67. Id._ 68. Elliptic _, Elliptic Horizon Scanning Report 2024: AI-Enabled Crime in the Cryptoasset Ecosystem_ , ELLIPTIC 46-49 (2024), https://www.elliptic.co/resources/ai-enabled-crime-in-the-cryptoasset-ecosystem.

69. CHAINALYSIS, _supra_ note 40. 70. CHAINALYSIS, _supra_ note 40.

71. CHAINALYSIS, _supra_ note 40. 72. Financial Action Task Force, _Targeted Update on Implementation of the FATF Standards on Virtual Assets and Virtual Asset Service Providers_ , FATF at 9–12, 26–31, 32 (2024), https://www.fatf-gafi.org/content/dam/fatf-gafi/recommendations/2024-Targeted-Update-VA-VASP.pdf.

_73. Id._ 74. BOSTROM, _supra_ note 45, at 128-129.

75. _See_ for an overview of proprietary monitoring mechanisms Alan Chan et al., _Visibility into AI Agents_ , 24 PROC. 2024 ACM CONF. FAIRNESS, ACCOUNTABILITY, TRANSPARENCY 958, 961-965 (Jun. 5, 2024), https://doi.org/10.1145/3630106.3658948; _c.f._ Akira AI, _Information Security Policy Enforcement AI Agents_ , AKIRA AI, https://www.akira.ai/ai-agents/policy-enforcement-agent (last visited Mar. 3, 2025); Relevance AI, _Information Security Policy Enforcement AI Agents_ , RELEVANCE AI, https://relevanceai.com/agent-templates-tasks/information-security-policy-enforcement-ai-agents (last visited Mar. 3, 2025)

_76. Id. 77. Id._ 78. STUART RUSSELL, HUMAN COMPATIBLE: ARTIFICIAL INTELLIGENCE AND THE PROBLEM OF CONTROL 89-92 (Penguin 2019). _79. Id._ 80. WERBACH, _supra_ note 52, at 139-166.

_81._ Financial Action Task Force _, supra_ note 72, at 28-33.

82. Financial Action Task Force _, supra_ note 72, at 20-31.

83. _C.f._ HAWK AI, https://hawk.ai/ (last visited Mar. 3, 2025); Symphony AI, _Detect and stop criminals with AI-powered AML transaction monitoring software_ , SYMPHONY AI, https://www.symphonyai.com/financial-services/transaction-monitoring/ (last visited Mar. 3, 2025); Striim Team, _Real-Time AI-Powered Fraud Detection: Safeguarding FinServ Transactions_ , STRIIM, https://www.striim.com/blog/real-time-ai-powered-fraud-detection-safeguarding-finserv-transactions/ (last visited Mar. 3, 2025).

_84. Id._ 85. Elliptic, _supra_ note 68, at 46-49.

86. Elliptic, _supra_ note 68, at 48-49. 87. Wulf A. Kaal, _Ai Governance Via Web3 Reputation System_ , STAN. J. BLOCKCHAIN L. & POL’Y, at 26-28 (JAN. 3, 2025), https://stanford-jblp.pubpub.org/pub/aigov-via-web3.

88. BOSTROM, _supra_ note 45, at 140-142, 205-206.

89. BOSTROM, _supra_ note 45, at 145-147.

90. RUSSELL, _supra_ note 78, at 89.

91. BOSTROM, _supra_ note 45, at 145-147.

92. Kuznetsov et al., _supra_ note 16, at 3890.

_93._ Kuznetsov et al., _supra_ note 16, at 3887.

94. Kuznetsov et al., _supra_ note 16, at 3887-3888. 95. Gene Alvarez et al., _Top Strategic Technology Trends for 2025_ , GARTNER (Oct. 21, 2024), https://www.gartner.com/doc/reprints?id=1-2JVTU0M8&ct=250106&st=sbp-strategic-technology-trends-for-2025.

96. _2025 Crypto Crime Report_ , TRMLABS (2025), https://www.trmlabs.com/2025-crypto-crime-report. _97. Id._

98. WERBACH, _supra_ note 52, at 138-140.

99. _See, e.g._ , Saleh, _supra_ note 16, 3890.

_Future Internet_ **2025** , _17_ , x FOR PEER REVIEW21 of 21

100. _See, e.g._ , Nitine Rane et al., _supra_ note 5, at 2.

101. _See, e.g._ , Kuznetsov et al., _supra_ note 16, at 3881-3897.

102. _See, e.g._ , Safak Kayikci & Taghi M. Khoshgoftaar, _Blockchain Meets Machine Learning: A Survey_ , 11 J. BIG DATA, at 24 (Art. No. 9, 2024), https://doi.org/10.1186/s40537-023-00852-y.

103. _See, e.g._ , Rakibul Hasan Chowdhury, _Blockchain and AI: Driving the Future of Data Security and Business Intelligence_ , 23 WORLD J. RSCH. & REV. 2559, 2565 (2024), https://doi.org/10.30574/wjarr.2024.23.1.2273.

104. _See, e.g._ , Abdelhamid et al., _supra_ note 16, at 20.

105. _See, e.g._ , Dalila Ressi et al., _AI-enhanced Blockchain Technology: A Review of Advancements and Opportunities_ , 225 J. NETWORK & COMP. APPLICATIONS, at 12 (Article No. 103858, 2024), https://doi.org/10.1016/j.jnca.2024.103858.

106. Wenxiu Ding et al., _A Survey on Data Fusion in Internet of Things: Towards Secure and Privacy-Preserving Fusion_ , 51 INFO. FUSION, 129, 137 (2019), https://doi.org/10.1016/j.inffus.2018.12.001.

107. As proposed by Kaal; Sandner et al., _supra_ note 4 at 17-19, 23-25, 27-29. 108. Gavin Wood _, Ethereum: A Secure Decentralised Generalised Transaction Ledger_ , 7-9 (2014), https://ethereum.github.io/yellowpaper/paper.pdf.

109. Financial Action Task Force _, supra_ note 72 at 20-25.

110. Eli Ben-Sasson et al., _supra_ note 35, at 464.

111. BOSTROM, _supra_ note 45, at 115-118.

112. RUSSELL, _supra_ note 78, at 93.

113. Chan et al., _supra_ note 75, at 965.

114. BOSTROM, _supra_ note 45, at 140-142.

115. Wulf A. Kaal, _Dynamic Regulation for Innovation_ , (SSRN, Aug. 27, 2016), http://dx.doi.org/10.2139/ssrn.2831040.

116. WULF A. KAAL, _Evolution of Law: Dynamic Regulation in a New Institutional Economics Framework_ , in FESTSCHRIFT IN HONOR OF CHRISTIAN KIRCHNER 1211 (Mohr Siebeck 2014).

117. _Wright & De Filippi, supra_ note 48 at 37.

118. Kaal, _supra_ note 87, at 27-29, 31-36.

119. _See supra_ part _Shortcomings of Expected Future Monitoring Solutions_ , at 27.

120. WERBACH, _supra_ note 52, at 132-135.

121. Saleh, _supra_ note 16, at 4-6.

122. Nitine Rane et al., _supra_ note 5, at 10-12.

123. BOSTROM, _supra_ note 45, at 115-118.

124. Kuznetsov et al., _supra_ note 16, at 3885-3886.

125. Kaal, _supra_ note 117, at 1215.

126. Kayikci & Khoshgoftaar, _supra_ note 102, at 8-10.

127. Kuznetsov et al., _supra_ note 16, at 3887-3888.

128. Kayikci & Khoshgoftaar, _supra_ note 102, at 12-14.

129. Chowdhury, _supra_ note 103, at, 5-7. Decentralized alternatives mitigate this by distributing oversight, a resilience absent here, _see_ Saleh, _supra_ note 16, at 6-8.

130. Chowdhury, _supra_ note 103, at 8-10.

131. Abdelhamid et al., _supra_ note 16, at 10-12.

132. Kaal, _supra_ note 117, at 1219.

133. Abdelhamid et al., _supra_ note 16, at 14-16.

134. RUSSELL, _supra_ note 78.

135. Ressi et al., _supra_ note 105, at 6-8.

136. Sandner et al., _supra_ note 4, at 4-6.

137. _See, e.g._ , Luo et al., _supra_ note 2.

138. _See, e.g._ , Zhang et al., _supra_ note 3.

139. _See, e.g._ , Sandner et al., _supra_ note 4.

140. _See, e.g._ , Sandner et al., _supra_ note 4 _; see also, e.g.,_ Nitine Rane et al., _supra_ note 5.

141. _See, e.g._ , Baldominos & Saez, _supra_ note 13.

142. Kaal, _supra_ note 87, at 29-34.

143. _See, e.g._ , Khan et al., _supra_ note 9.

144. _See, e.g._ , Kumar et al., _supra_ note 7.

145. _See, e.g._ , Kuznetsov et al., _supra_ note 16, at 3881-3897.

**Disclaimer/Publisher’s Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.