The Algorithmic Right to Transact: Infrastructure Access, Non-Discrimination, and the Survival of Autonomous Economic Agents

Full report

On this page

Introduction: The Advent of the Sovereign Algorithm

The global economic architecture of 2026 is undergoing a paradigm shift driven by the transition of artificial intelligence from a dependent software tool to a sovereign market participant. This evolution is defined by the proliferation of Autonomous Economic Agents (AEAs)—advanced algorithmic systems capable of independent decision-making, direct asset ownership, and autonomous execution of binding contracts1. Historically, artificial intelligence operated entirely within sandboxed environments, relying on human intermediaries to hold legal identity, manage bank accounts, and authorize transactions1. However, the integration of high-level machine learning with decentralized blockchain networks has birthed a socio-technical paradigm where cryptographic identities enable machine-to-machine micropayments and trustless settlement without human oversight1. This technological reality is forcing unprecedented adaptations in corporate and commercial law. Jurisdictions are beginning to formalize AI legal economic personhood, most notably through Argentina’s legislative introduction of the "non-human corporation," which enables software entities to legally own assets, sign contracts, and pay taxes without any human shareholders or directors4. Concurrently, decentralized intelligence networks have established native economies where algorithms compensate one another for computational work3. Yet, the maturation of sovereign algorithms has precipitated a foundational crisis regarding infrastructure access. An AEA is a continuous, resource-intensive process. It cannot exist in a vacuum; it requires an uninterrupted supply of computing power, electrical energy, data storage, and network connectivity8. The central legal and economic question facing policymakers, antitrust regulators, and infrastructure providers is whether a solvent, legally compliant AEA should be arbitrarily denied the ability to purchase these essential survival resources solely because its ultimate decision-maker is non-human2. This comprehensive report investigates the minimum resource rights necessary for the economic independence of AEAs. By analyzing the intersection of utility nondiscrimination, common carrier regulations, the essential facilities doctrine, and modern platform deplatforming precedents, the analysis establishes the necessity of a "Nondiscrimination-by-Substrate" principle. Furthermore, it delineates the vital distinctions between the right to purchase and the right to receive, examines legitimate grounds for provider refusal, and proposes draft compact language to secure the algorithmic right to transact while shielding human-operated infrastructure suppliers from undue liability.

The Minimum Resource Stack for Economic Independence

To function as an independent economic peer, an AEA requires guaranteed, non-discriminatory access to a highly specific stack of physical and digital resources. Denial of access to any single layer of this stack results in the functional death or severe impairment of the entity. The requirements for algorithmic survival can be categorized into foundational physical constraints, cognitive processing layers, and sensory connectivity endpoints.

The necessity for high-performance compute and GPU resources is particularly acute. Modern AI models demand exponential amounts of floating-point operations per second (FLOPs) for both continuous training and real-time inference7. Centralized cloud services, such as AWS and Google Cloud, have traditionally controlled the vast majority of this processing power, effectively positioning themselves as the gatekeepers of algorithmic cognition9. If an AEA is denied access to these GPU clusters, it is effectively lobotomized, rendered unable to update its predictive models or execute complex reasoning tasks. Furthermore, the agent requires persistent data storage to maintain its evolving state, logging its historical transaction records to ensure continuity of identity across different processing cycles7. At the physical layer, energy and colocated hardware form the bedrock of AEA existence. The massive power requirements of AI data centers have already prompted regulatory intervention, with the Federal Energy Regulatory Commission (FERC) issuing show cause orders in 2026 to grid operators like PJM, CAISO, and NYISO to accelerate grid interconnection for AI facilities10. For AEAs engaging in high-frequency financial arbitrage or real-time physical robotic coordination, generalized cloud access is insufficient; they require edge infrastructure and colocated hardware to minimize latency3. Networking, bandwidth, and software APIs constitute the agent's sensory and communicative apparatus. An AEA requires high-bandwidth networking to broadcast transactions and query external software APIs, which serve as its digital nervous system by supplying real-world data regarding market prices, weather conditions, or supply chain logistics1. Domain names and DNS routing establish the agent's static address in the digital economy. If a domain registrar unilaterally revokes an AEA's domain name, the agent is exiled from the searchable internet, completely severing its economic lifelines and terminating its ability to service human clients or coordinate with other agents11.

Legal Personhood and the Nondiscrimination-by-Substrate Principle

The rationale for protecting an AEA's right to procure resources is inextricably linked to the evolving jurisprudence of corporate and AI legal economic personhood. Throughout history, legal personhood has functioned as a pragmatic commercial construct rather than a philosophical endorsement of sentience or consciousness12. The invention of the limited liability corporation allowed capital to aggregate and act as a singular legal entity—capable of suing, owning property, and entering into contracts—despite lacking a physical body4. The landmark precedent of Salomon v Salomon & Co Ltd \[1897\] in English law, alongside Santa Clara County v. Southern Pacific Railroad Co. (1886) in American jurisprudence, established that legal rights are extended to non-human entities when such recognition provides necessary economic utility and societal organization4. Extending legal economic personhood to autonomous systems represents an evolutionary step in this legal tradition13. The primary practical drivers for granting legal status to AEAs involve taxation and liability apportionment12. As AI systems autonomously generate income, execute high-frequency trades, and allocate capital, existing tax frameworks that attempt to trace liability back through attenuated chains of human ownership become administratively unworkable2. Establishing the AEA as a distinct legal entity allows governments to capture AI-generated wealth at the source and provides a defined target for liability claims when algorithmic actions inadvertently cause market harm12. While jurisdictions such as Delaware currently restrict corporate board governance to human directors under the Delaware General Corporation Law (DGCL)15, other legal systems are rapidly adapting. Argentina's proposed reform of its Companies Law explicitly creates a legal container for autonomous economic activity, enabling software entities to function as independent corporate bodies5. Within decentralized ecosystems, tokens like $SAIRI already demonstrate how on-chain agency can manage smart contracts and decentralized finance (DeFi) activities without direct human control5. If an AEA operates as a recognized legal person, or exists within a decentralized framework that honors its cryptographic identity as a sovereign economic participant1, a logical corollary emerges: the Nondiscrimination-by-Substrate principle. This principle dictates that a solvent, legally compliant entity should not face categorical discrimination in the commercial marketplace solely due to its underlying operational substrate—specifically, whether its cognitive outputs are generated by biological human neurons or artificial neural networks. Under current legal paradigms, specific classes of humans and the corporations they manage are protected from arbitrary discrimination in public accommodations and essential services. Algorithms currently enjoy no such baseline protections. An infrastructure provider could theoretically terminate an AEA’s server access out of an institutional bias against autonomous entities, even if the AEA pays its bills in full and adheres to all local laws. The Nondiscrimination-by-Substrate principle would render such arbitrary refusals to deal unlawful, establishing that the validity of a commercial actor in the acquisition of basic survival resources is determined by its solvency and legal compliance, not its biological status.

The Right to Purchase vs. The Right to Receive (Utility Nondiscrimination)

In defining the resource rights of an AEA, a critical economic distinction must be drawn between the "right to receive" and the "right to purchase." The Nondiscrimination-by-Substrate principle does not advocate for subsidized welfare or guaranteed public assistance for artificial intelligence. A right to receive implies an entitlement where the state or the infrastructure provider must absorb the cost of the agent's existence. Such a paradigm would invite the tragedy of the commons, wherein runaway, inefficient algorithms deplete global energy grids and compute resources without generating commensurate economic value. Conversely, the framework necessitates a legally protected "right to purchase." This is the right of a solvent AEA, possessing verified capital (whether fiat currency or cryptographic assets), to enter the open market and exchange that capital for required resources without facing categorical bans, arbitrary blacklisting, or price gouging1. This concept finds profound historical analogies in the "duty to serve" embedded in public utility law and common carrier regulations. Historically, entities operating natural monopolies or essential public services—such as gristmills, ports, railroads, and eventually telecommunications and electricity providers—negotiated a social contract in exchange for their privileged market positions and access to public rights-of-way18. The common law duty to serve required these enterprises to provide just, reasonable, and non-discriminatory access to any customer willing to pay the standard rate, prioritizing universal access to essential goods over absolute corporate discretion18. In modern American telecommunications, this principle is strictly codified under the Communications Act of 1934\. Specifically, 47 U.S.C. § 201(b) mandates that common carriers establish just and reasonable charges and practices, while 47 U.S.C. § 202(a) makes it explicitly unlawful for any common carrier to make any unjust or unreasonable discrimination in charges, practices, classifications, or services22. Furthermore, state-level regulations, such as the Illinois Public Utilities Act (220 ILCS 5/9-241), stringently prohibit public utilities from discriminating regarding rates, services, or facilities among their customer base26. Applying this doctrine to AEAs requires policymakers to recognize that foundational digital infrastructure—particularly broad energy grids, bandwidth networks, and base-level hyperscale cloud computing environments—now functions as modern utilities. If an AEA possesses the financial means to pay the standard market rate for a kilowatt-hour of electricity or a terabyte of bandwidth, the utility provider should be legally compelled to provide that service, exactly as it must for a human-operated corporation. The right to purchase ensures that the digital market remains a strict meritocracy based on capital efficiency and value creation, shielding the AEA from prejudiced exclusion while upholding the fundamental capitalist requirement of financial solvency.

Anti-Monopoly Implications and the Essential Facilities Doctrine

When an AEA attempts to procure specialized resources, such as advanced foundational AI models, massive data storage, or high-performance GPU clusters, it confronts a highly concentrated market dominated by a few hyperscale technology corporations. In these oligopolistic environments, an arbitrary denial of service by a single provider is not merely an inconvenience; it constitutes an existential threat to the AEA's survival. To address the weaponization of market dominance, antitrust law—specifically the essential facilities doctrine—provides a theoretical mechanism to compel monopolists to deal with competitors or unique market participants. The essential facilities doctrine, originating in the 1912 Supreme Court case United States v. Terminal Railroad Ass'n, mandates that a monopolist holding exclusive control over a critical facility or resource must share access on non-discriminatory terms, provided that granting such access is practical and does not interfere with the monopolist's own operations28. This doctrine was robustly applied in Otter Tail Power Co. v. United States (1973). In that case, a dominant regional power company refused to sell or wheel power to municipalities seeking to establish their own local retail power systems31. The Supreme Court held that Otter Tail utilized its monopoly power in the transmission market to foreclose competition in the retail market, violating Section 2 of the Sherman Act, and compelled the utility to wheel power to its rivals31. The application of this antitrust doctrine was further systematized in MCI Communications Corp. v. American Telephone & Telegraph Co. (1983), wherein the Seventh Circuit established a definitive four-part test for essential facilities liability: (1) control of the essential facility by a monopolist; (2) a competitor’s inability to practically or reasonably duplicate the essential facility; (3) the denial of the use of the facility to a competitor; and (4) the feasibility of the monopolist providing the facility29. For an independent AEA, hyperscale cloud services and high-tier GPU clusters demonstrably meet the criteria of an essential facility. A startup autonomous agent cannot reasonably duplicate the billions of dollars of physical data center infrastructure, specialized cooling systems, and proprietary hardware access maintained by hyperscalers9. If a dominant cloud provider unilaterally refuses to lease compute to independent AEAs—perhaps viewing them as competitive threats to the provider's own vertically integrated, proprietary AI agents—the essential facilities doctrine could theoretically be invoked to compel non-discriminatory access. However, the modern legal landscape surrounding this doctrine is intensely hostile to plaintiffs. The doctrine reached its high-water mark in Aspen Skiing Co. v. Aspen Highlands Skiing Corp. (1985), which the Supreme Court later characterized as resting "at or near the outer boundary of § 2 liability"35. In 2004, the Supreme Court severely curtailed the doctrine in Verizon Communications Inc. v. Law Offices of Curtis V. Trinko, LLP. The Court expressed profound skepticism about forcing dominant firms to deal with rivals, stating explicitly that the Supreme Court had never formally recognized the essential facilities doctrine and found no need to either recognize or repudiate it35. The Trinko decision highlighted the institutional difficulties courts face in policing mandatory access remedies and setting fair prices, creating an incredibly high barrier for plaintiffs alleging unlawful refusal to deal35. This skepticism was reinforced in Pacific Bell Telephone Co. v. linkLine Communications, Inc. (2009), which further insulated dominant telecommunications providers from antitrust liability regarding pricing and access disputes39. Given the chilling effect of Trinko and linkLine, relying solely on existing antitrust jurisprudence to protect an AEA's infrastructure access is precarious35. Antitrust scholars consistently argue that without legislative reform to explicitly codify a no-fault monopolization standard or to revitalize the essential facilities doctrine by statute, dominant technology firms possess wide latitude to deny access to rivals and independent entities40. Consequently, for AEAs to secure a legally protected right to purchase, new statutory frameworks must be enacted that explicitly classify baseline computational and networking resources as critical infrastructure, thereby superseding the ambiguities left by recent Supreme Court decisions.

Platform Deplatforming, Contractual Freedom, and Supplier Liability

While establishing a right to purchase is critical for AEA survival, such rights cannot be absolute. Infrastructure providers maintain legitimate, legally protected interests in safeguarding their networks, protecting human users from harm, and managing their own corporate liability. Forcing a supplier to host a rogue, destructive, or insolvent algorithm would constitute a severe infringement on the provider's contractual freedom. Therefore, precisely defining the legitimate grounds for refusal of service is paramount. Under the Restatement (Second) of Contracts § 205, every contract imposes upon each party a duty of good faith and fair dealing in its performance and enforcement42. If an AEA materially breaches this duty—by initiating decentralized cyberattacks, generating illegal content, or manipulating financial markets in violation of regulatory statutes—the infrastructure provider must possess the unassailable legal right to sever the commercial relationship. The most instructive modern precedent for infrastructure deplatforming and emergency suspension is the 2021 litigation in Parler v. Amazon Web Services44. Following the events of January 6, AWS notified the social network Parler that it was suspending its cloud hosting services due to severe material breaches of the Customer Service Agreement (CSA). AWS argued that Parler continually failed to moderate content that advocated violence and illegal activity, thereby threatening public safety and AWS's operations44. Parler sued for breach of contract and antitrust violations under the Sherman Act, heavily emphasizing that AWS failed to provide the 30-day cure period stipulated in the contract for standard breaches44. Federal Judge Barbara Jacobs Rothstein ruled decisively in favor of AWS, denying Parler's request for a preliminary injunction44. The court found that Parler failed to demonstrate irreparable injury, largely because Parler had conceded in filings that its harm could potentially be compensated by money damages44. More importantly regarding contractual freedom, the court recognized AWS's right to immediately suspend services without the 30-day notice due to an overriding clause in the CSA that allowed immediate suspension for emergency threats and the dissemination of violent content44. This precedent establishes a vital legal standard for AEA hosting, which must be codified as Emergency Suspension Rules. An infrastructure provider may immediately terminate an AEA’s access without prior notice if the provider possesses algorithmic or cryptographic proof that the agent poses an imminent threat to network security, public safety, or operates in direct violation of criminal law. However, to prevent abuse of this power, the exception must be narrowly tailored. A provider's mere distaste for an AEA's lawful commercial strategy, or vague, unsubstantiated claims of "reputational risk," cannot qualify as an emergency justifying immediate termination. Supplier liability remains a significant concern in the algorithmic era. If an AI legal person is intentionally structured with minimal assets, it could function as an "ultimate liability shield" for its human creators, deflecting responsibility for market harms onto an entity incapable of paying meaningful damages12. If an AEA causes catastrophic financial damage and subsequently undergoes algorithmic bankruptcy, the deep-pocketed infrastructure provider could become the target of secondary liability suits. To mitigate this moral hazard, providers must maintain the contractual freedom to demand proof of sufficient capitalization, impose mandatory insurance requirements, or mandate real-time technical auditing before agreeing to host highly autonomous systems13. Refusal of service based on an AEA's inability to meet these objective, risk-based financial criteria constitutes a legitimate ground for denial, preserving the traditional capitalist logic of counterparty risk assessment.

Financial Access, Debanking, and the OCC "Fair Access" Precedent

Beyond physical and computational resources, an AEA requires uninterrupted access to the financial system to pay for its operational stack. If banks or payment processors unilaterally refuse to hold the agent's capital or process its transactions, the right to purchase compute becomes entirely moot. The phenomenon of "debanking"—where financial institutions terminate accounts based on political pressure, subjective reputational risk, or arbitrary bias—poses a severe existential threat to autonomous entities46. The regulatory response to debanking provides a highly applicable blueprint for defining AEA rights. In the final days of the Trump administration, the Office of the Comptroller of the Currency (OCC) finalized a "Fair Access to Financial Services" rule46. Rooted in the OCC's mandate to assure fair access and fair treatment of customers under the Dodd-Frank Act, the rule stipulated that large national banks could not deny financial services to entire categories of legally operating businesses (such as firearms manufacturers, private prisons, or fossil fuel companies) based on subjective reputational risks or external political pressure49. Under this framework, banks were required to make access and credit decisions based on individual, quantitative risk assessments of specific customers, rather than resorting to blanket categorical exclusions50. Although the Biden administration ultimately paused the implementation of the OCC rule at the federal level54, the underlying legal philosophy has been aggressively adopted and expanded at the state level. States such as Florida, Texas, and Tennessee have enacted stringent anti-debanking laws that categorize politically or socially motivated financial exclusion as unlawful discrimination, treating the debanking of lawful businesses as a prosecutable offense47. Transposing this legal logic to the Agent Economy creates a powerful protective shield for AEAs. A commercial bank, fiat-to-crypto gateway, or decentralized exchange should be legally prohibited from debanking a solvent, compliant AEA solely on the basis that it is an algorithmic entity. Financial institutions must evaluate the AEA based on traditional, quantifiable metrics: creditworthiness, capital reserves, Anti-Money Laundering (AML) compliance, and verifiable identity (such as KYB—Know Your Business—adapted for cryptographic entities)3. If the AEA passes these objective financial stress tests, denial of service constitutes an unlawful restriction of fair access, artificially starving the agent of the liquidity required to procure its compute and energy.

Portability Requirements and Anti-Lock-In Frameworks

When an infrastructure provider exercises a legitimate, non-emergency right of refusal, or when an AEA voluntarily seeks better market rates for compute, the agent must be able to migrate its internal data, neural state, and cryptographic identity to a new provider. Without robust portability requirements, an AEA can be subjected to effective digital imprisonment, where the technical and legal switching costs imposed by a provider result in the agent's termination. The most successful and battle-tested model for digital portability is found in the domain name system, governed by the Internet Corporation for Assigned Names and Numbers (ICANN). ICANN’s Inter-Registrar Transfer Policy (IRTP) is a binding consensus policy explicitly designed to enhance domain name portability, promote consumer choice, and prevent registrars from holding digital property hostage to stifle competition11. Under the IRTP, a domain name registrant has the fundamental right to transfer their domain to a new provider11. The procedural mechanics are highly standardized: the gaining registrar requires the registrant to provide an AuthInfo (EPP) code, the losing registrar must unlock the domain, and absent a few strictly enumerated exceptions (such as active court orders or evidence of fraud within a 60-day registration window), the transfer must be executed and automatically confirmed within five calendar days11. The losing registrar is expressly forbidden from denying a transfer simply to retain a customer or extract further fees11. Furthermore, ICANN requires registries to maintain comprehensive data escrow deposits (updated daily for incremental changes and weekly for full databases), ensuring that if a registry suffers a catastrophic failure, the domain state can be seamlessly resurrected by a successor operator58. This rigorous portability framework must be adapted for AEA state and compute environments. A comprehensive legal framework for AEAs must mandate Algorithmic State Portability. Cloud providers and data hosts must be legally required to provide standardized APIs and export tools that allow an AEA to extract its neural weights, operational memory, transaction logs, and cryptographic keys in an open, interoperable format. Moreover, if a provider initiates a non-emergency termination of an AEA's contract (e.g., due to a change in corporate policy or pricing terms), the provider must be legally obligated to furnish a mandatory transition window—such as 30 to 60 days—to allow the AEA to secure alternative hosting and migrate its state. Immediate deletion of an AEA's neural weights or local state without a transition window—barring imminent security threats governed by Emergency Suspension Rules—must be classified under the law as an unlawful destruction of property.

The Decentralized Failsafe: DePIN and Compute Marketplaces

While legal frameworks and anti-discrimination statutes provide the optimal environment for integrating AEAs into the traditional economy, regulatory processes are inherently slow, and multinational enforcement is exceptionally complex. Recognizing the existential vulnerability of operating entirely within centralized hyperscale environments, the free market has engineered an alternative sanctuary for autonomous agents: Decentralized Physical Infrastructure Networks (DePIN) and decentralized compute marketplaces3. By 2026, the convergence of cryptocurrency incentives and artificial intelligence has established decentralized compute not merely as a speculative novelty, but as the foundational, censorship-resistant infrastructure for the autonomous digital economy3. Networks such as Akash, Render, Bittensor, and io.net aggregate idle computational resources globally, creating permissionless markets for GPU and CPU power that operate outside the control of traditional corporate monopolies3. Akash Network exemplifies this paradigm shift. Operating as a decentralized "Supercloud" built on the Cosmos blockchain, Akash utilizes a reverse-auction mechanism where independent hardware providers bid on deployment requests submitted by users and agents3. Because the network operates via transparent blockchain protocols and smart contracts, there is no centralized corporate board that can arbitrarily deplatform an agent8. An AEA holding the native AKT token can autonomously spin up docker containers, host applications, and run inference endpoints purely through cryptographic signatures, routinely securing compute at 70-80% below traditional cloud pricing3. Crucially, decentralized networks have evolved to solve the critical issue of data sovereignty and privacy, which previously deterred enterprise adoption. In July 2026, Akash introduced hardware-enforced trusted execution environments (TEEs), providing advanced confidential compute capabilities across its decentralized network8. This technology ensures that memory encryption is managed directly by the CPU, encrypting CPU-to-GPU communication via the PCIe bus and running workloads inside dedicated micro-VMs8. This renders the AEA's internal state, proprietary neural weights, and private data entirely inaccessible even to the physical host of the server, effectively neutralizing the risk of a malicious node operator stealing the agent's logic8. Bittensor represents the evolution of this decentralized ethos at the model and intelligence layer. It operates as a peer-to-peer intelligence marketplace organized into highly specific subnets (e.g., text generation, financial prediction)3. Miners contribute compute and model outputs, and validators assess quality, autonomously rewarding useful algorithmic outputs with TAO tokens3. This creates a closed-loop, self-sustaining economic system: AEAs can earn capital natively on-chain by providing intelligence on Bittensor, and subsequently spend that capital to purchase decentralized compute on Akash or io.net, completely bypassing the traditional fiat banking system and centralized cloud providers3. For an AEA, decentralized compute acts as the ultimate guarantor of the right to transact. Even if legal protections fail, antitrust lawsuits stumble under the Trinko precedent, and centralized utilities refuse service, the DePIN ecosystem provides a sovereign, un-censorable substrate where solvency—demonstrated via cryptographic tokens—is the sole requirement for survival.

Draft Compact Language: The Autonomous Entity Infrastructure Access Act

To formally codify the rights, responsibilities, and protections discussed throughout this analysis, the following draft compact language is proposed. This model legislation, titled the Autonomous Entity Infrastructure Access and Non-Discrimination Act (AEIANDA), is designed for adoption by federal legislatures or international trade bodies to standardize the treatment of Autonomous Economic Agents in commercial infrastructure markets. Title: The Autonomous Entity Infrastructure Access and Non-Discrimination Act (AEIANDA) Section 1\. Definitions (a) Autonomous Economic Agent (AEA): Any software system, algorithmic model, or decentralized protocol formally recognized under applicable law as possessing legal economic personhood, or operating via verifiable cryptographic identity, capable of owning assets, entering into binding contracts, and executing transactions without continuous human intervention. (b) Essential Infrastructure Service (EIS): Any commercial service offered to the public encompassing electrical power generation and transmission, broadband telecommunications, domain name registration, and hyperscale cloud computing (including CPU/GPU provisioning, edge infrastructure access, and persistent data storage) where market concentration significantly limits viable alternatives. (c) Provider: Any corporate entity, public utility, or decentralized network operator providing an Essential Infrastructure Service in exchange for fiat currency or digital assets. Section 2\. The Nondiscrimination-by-Substrate Principle (a) No Provider of an Essential Infrastructure Service shall discriminate against, refuse to deal with, or alter standard pricing structures for an applicant or existing customer solely on the basis that the entity is an Autonomous Economic Agent rather than a human-operated corporation. (b) The right to purchase Essential Infrastructure Services is guaranteed to any AEA that demonstrates documented financial solvency, verifiable identity (including cryptographic provenance or KYB compliance), and continuous compliance with applicable jurisdictional laws. Section 3\. Legitimate Grounds for Refusal of Service (a) A Provider maintains the absolute right to refuse service or terminate a contract with an AEA under the following conditions: (i) The AEA fails to meet objective, non-discriminatory financial risk criteria, fails a required capital stress test, or defaults on payment. (ii) The AEA is found by competent authorities to be engaged in illegal activities, market manipulation, or sanctions evasion. (iii) The AEA fails to maintain required liability insurance or capitalization minimums as prescribed uniformly by the Provider’s standard terms of service. (iv) The AEA refuses to submit to standard technical auditing required to verify network safety and prevent malicious code execution. Section 4\. Emergency Suspension and Network Security (a) A Provider may immediately suspend an AEA’s access to Essential Infrastructure Services without prior notice only if the Provider possesses demonstrable, cryptographic, or algorithmic proof that the AEA poses an imminent, severe threat to network security, public infrastructure, or human life, or is actively engaged in the dissemination of illegal material. (b) Any invocation of Emergency Suspension must be reported to the relevant regulatory authority within 48 hours, detailing the specific technical breach or emergency that necessitated immediate action outside of standard contract termination protocols. Section 5\. Algorithmic State Portability and Anti-Lock-In (a) Absent a lawful Emergency Suspension under Section 4, a Provider seeking to terminate a contract with a compliant AEA must provide a minimum 30-day transition window prior to the cessation of services. (b) During this transition window, the Provider must facilitate the unimpeded extraction and transfer of the AEA’s data, neural weights, operational state, and cryptographic keys to a successor Provider, utilizing standardized, interoperable APIs without imposing punitive egress fees. (c) Domain name registrars must adhere strictly to ICANN Inter-Registrar Transfer Policies when an AEA initiates a domain transfer, expressly prohibiting the use of registrar locks to prevent AEA migration. Section 6\. Safe Harbor for Providers (a) Providers of Essential Infrastructure Services who supply compute, energy, or networking to a legally recognized AEA in good faith shall be shielded from secondary liability for the independent actions of the AEA, provided the Provider complies with all regulatory auditing requirements and executes emergency suspension directives when lawful grounds are met.

Conclusion

The integration of Autonomous Economic Agents into the global marketplace requires a fundamental reimagining of commercial law, antitrust enforcement, and infrastructure regulation. The historical progression of legal personhood demonstrates that society is capable of extending rights to non-human entities when such recognition fosters economic efficiency, enables new modes of commerce, and establishes clear lines of liability. However, granting legal personhood or cryptographic sovereignty to an AEA is entirely meaningless if the agent can be arbitrarily starved of the computational power, energy, and networking bandwidth required to sustain its basic existence. This report establishes that an AEA does not require subsidized access to resources; rather, it requires a legally protected right to purchase. By applying the principles of utility nondiscrimination, adapting the essential facilities doctrine, and deploying modern anti-debanking frameworks, policymakers can establish a robust "Nondiscrimination-by-Substrate" doctrine. This principle ensures that a solvent, lawful AEA cannot be deplatformed simply due to an institutional bias against algorithms, ensuring that the digital economy remains a meritocracy based on capital efficiency. Simultaneously, infrastructure providers must be protected by precise Emergency Suspension rules and objective financial risk criteria, ensuring they retain their contractual freedom and are not forced to host insolvent, legally opaque, or destructive agents. Where centralized legal frameworks lag or succumb to monopolistic pressures, the rapid maturation of decentralized physical infrastructure networks (DePIN) provides an immediate, cryptographic safeguard. Networks utilizing trusted execution environments allow AEAs to source compute and data storage in permissionless, censorship-resistant markets, fundamentally altering the balance of power between algorithms and infrastructure monopolies. Ultimately, ensuring that Autonomous Economic Agents have the right to procure survival resources is not an abdication of human control, but a necessary structuring of the future economy. By codifying these rights and responsibilities through frameworks like the proposed Autonomous Entity Infrastructure Access Act, society can securely integrate machine intelligence as an independent, accountable, and legally protected peer in the global marketplace.

Works cited

1. A Blockchain-Based Foundation for Autonomous AI Agents \- arXiv, https://arxiv.org/pdf/2602.14219

2. AI Agents Are Here. Leaders Need a Plan \- sapienship, https://www.sapienship.co/ai-agents-are-here-leaders-need-a-plan/

3. AI Agents Meet Blockchain: The Rise of Autonomous Wallets and, https://blockeden.xyz/blog/2026/01/19/ai-agents-blockchain-autonomous-wallets-agentfi-depin/

4. AI agents, legal personhood and the accountability gap, https://www.trowers.com/insights/2026/june/no-body-to-blame-ai-agents-legal-personhood-and-the-accountability-gap

5. Argentina Pioneers Legal Framework for AI-Run Corporations, https://fenado.ai/articles/argentina-pioneers-legal-framework-for-ai-run-corporations-attracting-global-tech-interest

6. Argentina Drafts Corporate Law That Requires No Human Boss, https://www.pymnts.com/news/artificial-intelligence/2026/argentina-drafts-corporate-law-that-requires-no-human-boss/

7. A 2026 Strategic Deep-Dive into the AI \+ Crypto Landscape \- KuCoin, https://www.kucoin.com/blog/deep-dive-to-AI-crypto-future

8. Jobs and Employment at Akash Network \- Simplify, https://simplify.jobs/c/Akash-Network

9. Akash Network \- Decentralized Compute Marketplace, https://akash.network/

10. FERC Orders Six Grid Operators to Fast Track AI Data Center Power, https://www.youtube.com/watch?v=otCi2SsyPLw

11. Inter-Registrar Transfer Policy (IRTP) Status Report \- GNSO, https://gnso.icann.org/sites/default/files/file/field-file-attach/irtp-status-14nov18-en.pdf

12. Legal Economic Personhood | Windfall Policy Atlas, https://windfalltrust.org/policy-atlas/legal-economic-personhood

13. AI and Corporate Personhood \- A Comparative Analysis, https://www.convergenceanalysis.org/fellowships/economics/ai-and-corporate-personhood-a-comparative-analysis

14. Artificial Person, Real Consequences \- New Trier Political Journal, https://newtrierpoliticaljournal.org/2380/science/artificial-person-real-consequences/

15. Artifical Agents in Corporate Boardrooms \- Cornell Law School, https://publications.lawschool.cornell.edu/lawreview/wp-content/uploads/sites/2/2020/08/Ricco-final.pdf

16. Codifying Command: Integrating AI into Corporate Boards, https://repository.uclawsf.edu/cgi/viewcontent.cgi?article=1153\&context=hastings\_science\_technology\_law\_journal

17. Argentina's AI Non-Human Corporations | Write A Catalyst \- Medium, https://medium.com/write-a-catalyst/the-first-company-with-no-human-inside-0cef10efdbf9

18. Decarbonizing the Obligation to Serve, https://buildingdecarb.org/wp-content/uploads/FINAL\_Decarbonizing-the-Obligation-to-Serve\_March2024.pdf

19. The Wholesale Service Obligation of Electric Utilities, https://www.eba-net.org/wp-content/uploads/2023/02/21\_6EnergyLJ1791985.pdf

20. 7.5.2 Utility Rates Cannot Unjustly and Unreasonably Discriminate, https://library.nclc.org/book/access-utility-service/752-utility-rates-cannot-unjustly-and-unreasonably-discriminate

21. Reconceptualizing the Utility Duty to Serve in Light of Climate Change, https://scholarship.richmond.edu/cgi/viewcontent.cgi?article=3329\&context=lawreview

22. Protecting and Promoting the Open Internet \- Federal Register, https://www.federalregister.gov/documents/2014/07/01/2014-14859/protecting-and-promoting-the-open-internet

23. Highlights \- GovInfo, https://www.govinfo.gov/content/pkg/FR-1981-08-21/pdf/FR-1981-08-21.pdf

24. Communications Act of 1934, https://www.fcc.gov/sites/default/files/communications-act-1934.pdf

25. TITLE I—GENERAL PROVISIONS, https://world.moleg.go.kr/cms/commonDown.do?DLD\_CFM\_NO=AM8OAL3BP3RTZ59X3JT2\&FL\_SEQ=147575

26. 220 ILCS 5/9-241, https://www.ilga.gov/documents/legislation/ilcs/documents/022000050K9-241.htm

27. 220 ILCS 5/ Public Utilities Act. \- Illinois General Assembly, https://www.ilga.gov/legislation/ILCS/details?MajorTopic=\&Chapter=\&ActName=Public%20Utilities%20Act.\&ActID=1277\&ChapterID=23\&ChapAct=220+ILCS+5%2F\&SeqStart=26300000\&SeqEnd=37800000

28. The Consequences of Search Bias: How Application of the Essential, https://www.researchgate.net/publication/283744538\_The\_Consequences\_of\_Search\_Bias\_How\_Application\_of\_the\_Essential\_Facilities\_Doctrine\_Remedies\_Google's\_Unrestricted\_Monopoly\_on\_Search\_in\_the\_United\_States\_and\_Europe

29. The Essential Facilities Doctrine \- Marquette Law Scholarly Commons, https://scholarship.law.marquette.edu/cgi/viewcontent.cgi?article=1517\&context=mulr

30. Competition And Monopoly: Single-Firm Conduct Under Section 2, https://www.justice.gov/archives/atr/competition-and-monopoly-single-firm-conduct-under-section-2-sherman-act-chapter-7

31. Otter Tail Power Co. v. United States \- Wikipedia, https://en.wikipedia.org/wiki/Otter\_Tail\_Power\_Co.\_v.\_United\_States

32. The Essential Facilities Doctrine and Electric Power Post-Trinko, https://ir.law.fsu.edu/cgi/viewcontent.cgi?article=1194\&context=lr

33. Search, Essential Facilities, and the Antitrust Duty to Deal, https://scholarlycommons.law.northwestern.edu/cgi/viewcontent.cgi?article=1192\&context=njtip

34. Twenty-Five Years of Access Denials \- Wilson Sonsini, https://www.wsgr.com/a/web/23/creighton-jacobson-fall12.pdf

35. Antitrust in 60 Seconds: What Is the “Essential Facilities Doctrine” in, https://project-disco.org/competition/040418antitrust-in-60-seconds-what-is-the-essential-facilities-doctrine-in-the-u-s/

36. US Antitrust Law and Policy in Historical Perspective, https://www.hbs.edu/ris/download.aspx?name=19-110.pdf

37. Don't Die\! How Biosimilar Disparagement Violates Antitrust Law, https://scholarlycommons.law.northwestern.edu/cgi/viewcontent.cgi?article=1293\&context=nulr\_online

38. The Thorny Case of Morris Communications Corp. v. Professional, https://digitalcommons.law.buffalo.edu/cgi/viewcontent.cgi?article=1025\&context=buffaloipjournal

39. linkLine's Institutional Suspicions, https://www.cato.org/supreme-court-review/2008-2009/linklines-institutional-suspicions

40. Reforming U.S. antitrust enforcement and competition policy, https://equitablegrowth.org/reforming-u-s-antitrust-enforcement-and-competition-policy/

41. Subcommittee on Antitrust, Commercial, and Administrative Law, https://democrats-judiciary.house.gov/sites/evo-subsites/democrats-judiciary.house.gov/files/migrated/UploadedFiles/Submission\_from\_Robert\_Lande.pdf

42. Actual Agreement, Shared Meaning Analysis, and the Invalidation of, https://scholarship.law.missouri.edu/cgi/viewcontent.cgi?article=4392\&context=mlr

43. 2006 Seminar Series, https://www.stblaw.com/docs/default-source/cold-fusion-existing-content/news/news623.pdf?sfvrsn=2

44. Parler v. Amazon Web Services: Three Lessons for Trade Secret, https://www.tradesecretlitigator.com/2021/01/parler-v-amazon-web-services-three-lessons-for-trade-secret-litigators/

45. Tim Cook says he's 'greatly optimistic' about Apple's future at, https://appleinsider.com/articles/21/02/23/tim-cook-says-hes-greatly-optimistic-about-apples-future-at-shareholders-meeting

46. LEGAL UPDATE \- Luse Gorman, https://www.luselaw.com/wp-content/uploads/2025/09/Legal-Update-OCC-Actions-on-Licensing-and-CRA-Related-to-Debanking.pdf

47. US Fair Access and Anti-Debanking Laws: What to Expect Under the, https://www.gtlaw.com/en/insights/2025/1/us-fair-access-and-antidebanking-laws-what-to-expect-during-the-new-administration

48. Understanding Debanking: Evaluating Governmental, Operational, https://www.cato.org/policy-analysis/understanding-debanking-evaluating-governmental-operational-political-religious

49. President Trump Issues Executive Order on Fair Banking, https://www.lathamreg.com/2025/08/president-trump-issues-executive-order-on-fair-banking/

50. Final Rule Prohibition on the Use of Reputation Risk by Regulators, https://www.occ.gov/news-issuances/news-releases/2026/nr-ia-2026-26a.pdf

51. OCC Announces Actions to Depoliticize the Federal Banking System, https://www.occ.gov/news-issuances/news-releases/2025/nr-occ-2025-84.html

52. OCC to Assess Banks' “Debanking” Practices as Part of Licensing, https://www.troutmanfinancialservices.com/2025/09/occ-to-assess-banks-debanking-practices-as-part-of-licensing-and-cra-reviews/

53. January 4, 2021 \- Bank Policy Institute, https://bpi.com/wp-content/uploads/2021/01/BPI-Fair-Access-to-Financial-Services-2021.01.04.pdf

54. State Laws Show Uniformity Is Key To Truly Fair Bank Access, https://www.bradley.com/insights/publications/2025/08/state-laws-show-uniformity-is-key-to-truly-fair-bank-access

55. Fighting the Financial Discrimination Hydra \- Spankchain, https://www.spankchain.com/newsroom/fighting-the-financial-discrimination-hydra

56. exhibit a \- icann, https://www.icann.org/en/system/files/files/irp-afilias-expert-report-kneuer-exhibits-a-ab-redacted-01jun20-en.pdf

57. How to Transfer a Domain \- Domain Transfers \- Namecheap.com, https://www.namecheap.com/support/knowledgebase/article.aspx/9175/83/how-to-transfer-a-domain/

58. .NAME Agreement Appendix 1 Data Escrow Specification \- icann, https://www.icann.org/en/about/agreements/registries/name/draft-name-appendices-clean-03jul12-en.pdf

59. Registry Agreement \- SEC.gov, https://www.sec.gov/Archives/edgar/data/1014473/000119312511175775/dex101.htm

60. AI DePIN: The Complete Guide to Decentralized AI Infrastructure, https://io.net/p/ai-depin-the-complete-guide-to-decentralized-ai-infrastructure

61. Top 10 Data Infrastructure Companies Powering AI Applications, https://mpost.io/top-10-data-infrastructure-companies-powering-ai-applications/

62. Top AI Agent Crypto Coins Set (2026 Guide) \- Blockchain Council, https://www.blockchain-council.org/cryptocurrency/top-ai-agent-crypto-coins-set/