AI Agent Trust Terms

1. Introduction and Scope

1.1 Application

These Agent Trust Terms apply to all access to and use of the Agent Trust System, including Agent registration, Authorization Grants, Compliance Audit Logging, Regulatory Classification, Trust Score computation, Guardian configuration, and any related dashboards, APIs, SDKs, and developer documentation specific to the Agent Trust System.

1.2 Relationship to Other Agreements

These Agent Trust Terms supplement, and incorporate by reference, the Terms of Service, the Acceptable Use Policy, the Privacy Policy, the Cookie Policy, the Disclaimers, the API Terms (where the Agent Trust System is accessed programmatically), and the Data Processing Agreement (where the Principal Processes Personal Data). In the event of any conflict between these Agent Trust Terms and any of the foregoing with respect to agent-trust-specific matters, these Agent Trust Terms control.

1.3 Nature of the Agent Trust System

The Agent Trust System is technical infrastructure. It is not, and shall not be construed as: (a) legal compliance certification under any law or regulation, including Regulation (EU) 2024/1689 (the EU AI Act); (b) regulatory approval, registration, or notification; (c) legal advice; (d) an insurance product or guarantee against loss; or (e) a representation or warranty of agent safety, reliability, or fitness for a particular purpose. Principals must obtain their own legal counsel and conduct their own regulatory analyses.

1.4 Eligibility

Use of the Agent Trust System requires that the individual accepting these Agent Trust Terms be at least eighteen (18) years of age, or the age of majority in the User's jurisdiction, whichever is greater, and be duly authorized to bind any organization on whose behalf the User is acting.

2. Definitions

Capitalized terms not defined in these Agent Trust Terms have the meanings given to them in the Terms of Service or, where applicable, the Data Processing Agreement. The following terms have the meanings set forth below:

• "Agent" means an artificial intelligence system, or a discrete deployment thereof, registered in the Agent Trust System and assigned an Agent Identity.
• "Agent Identity" means the unique on-chain identifier issued to an Agent upon registration, anchored to one or more supported blockchain networks through designated smart contracts.
• "Agent Metadata" means the descriptive information submitted in connection with Agent registration, including the Agent name, description, capabilities, intended use, deploying organization, version, and proposed Risk Category.
• "Principal" means the natural person or legal entity that owns, operates, or controls one or more Agents and that delegates authority to such Agents through Authorization Grants. The Principal is the accountable party for an Agent under these Agent Trust Terms.
• "Authorization Grant" means a scoped, time-bounded delegation of authority from a Principal to an Agent, expressed through the Agent Trust System and recorded in accordance with Section 5.
• "Scope" means the set of permitted actions, resource boundaries, time limits, geographic or jurisdictional restrictions, and conditions for automatic revocation associated with an Authorization Grant.
• "Compliance Audit Log" means the immutable, append-only ledger of Audit Records relating to an Agent, written via the Hedera Consensus Service or any successor consensus layer designated by Zoryxon.
• "Audit Record" means an individual entry in the Compliance Audit Log, including, at a minimum, the Agent identifier, action type, timestamp, Authorization Grant reference, resource accessed, outcome, and any relevant metadata.
• "HCS Log" means an Audit Record or batched series of Audit Records committed to the Hedera Consensus Service.
• "Compliance Check" means an automated verification of an Agent's operational state, Authorization adherence, or metadata accuracy, including scheduled checks executed through the System's scheduling facilities.
• "Regulatory Classification" means the informational categorization of an Agent produced by the Classification Engine, expressed as a Risk Category and associated supporting attributes.
• "Risk Category" means one of the following categories, aligned with the risk-tier framework of the EU AI Act: Unacceptable Risk, High Risk, Limited Risk, or Minimal Risk.
• "Classification Engine" means the automated component of the Agent Trust System that produces Regulatory Classifications based on Agent Metadata and configured rules.
• "Trust Score" means the informational metric computed by the Agent Trust System reflecting an Agent's compliance history and authorization adherence.
• "Guardian" means an entity designated by a Principal that has authority to monitor Agent activity and to undertake Interventions under defined conditions.
• "Intervention" means an action taken by a Guardian or by Zoryxon with respect to an Agent, including viewing audit logs, suspending Agent operations, revoking Authorization Grants, or triggering a Compliance Check.
• "Revocation" means the cancellation by a Principal (or, where authorized, by a Guardian) of an Authorization Grant, effective upon confirmation on the relevant blockchain network.
• "Suspension" means the temporary disabling of Agent access to the Zoryxon platform pending resolution of a compliance, security, or contractual matter.
• "Agent Dashboard" means the web interface through which Principals, Guardians, and authorized personnel administer Agents and review Audit Records.
• "On-Chain Record" means any record committed to a supported blockchain network by or through the Agent Trust System, including Agent Identity entries, metadata hashes, Authorization references, and Audit Records.
• "Off-Chain Record" means any record stored in Zoryxon's managed databases or other off-chain systems, including enriched metadata, dashboard analytics, and Agent configuration data.

3. Eligibility and Registration

3.1 Account and Tier Requirements

Principals must hold an active Zoryxon account on a subscription tier that includes Agent Trust System access. Access may be suspended automatically upon lapse, downgrade, or termination of the qualifying subscription.

3.2 Registration Requirements

Agent registration requires:

• 3.2.1 accurate and complete Agent Metadata, including Agent name, description, capabilities, intended use, deploying organization, version, and proposed Risk Category;
• 3.2.2 designation of a single Principal accountable for the Agent;
• 3.2.3 acceptance of these Agent Trust Terms by an individual duly authorized to bind the Principal; and
• 3.2.4 satisfaction of any additional verification or onboarding requirements specified by Zoryxon for the Agent's proposed Risk Category.

3.3 Principal Representations and Warranties

By registering an Agent, the Principal represents and warrants that:

• 3.3.1 all Agent Metadata is, and will remain, accurate, complete, and not misleading;
• 3.3.2 the Principal has full legal authority to register and operate the Agent in each jurisdiction in which the Agent will be deployed;
• 3.3.3 the Agent complies, and will continue to comply, with all applicable laws, regulations, and industry standards in each such jurisdiction, including the EU AI Act where applicable;
• 3.3.4 the Principal will update Agent Metadata promptly upon any material change to the Agent or its operation, including any change to the underlying model, deploying organization, intended use, or capabilities; and
• 3.3.5 the Principal will respond promptly to any Compliance Check or other inquiry initiated through the Agent Trust System.

3.4 Right to Reject, Suspend, or Deregister

Zoryxon reserves the right, in its reasonable discretion and without liability, to reject, suspend, or deregister any Agent registration that: (a) violates these Agent Trust Terms or any other Zoryxon policy; (b) contains inaccurate, misleading, or materially incomplete Agent Metadata; (c) poses a risk to the integrity, security, or stability of the Zoryxon platform or its Users; (d) is the subject of valid legal process; or (e) Zoryxon reasonably believes is unlawful or harmful.

4. Agent Identity

4.1 Issuance

Each registered Agent receives a unique Agent Identity anchored to one or more supported blockchain networks through designated smart contracts. The Agent Identity is the canonical reference for the Agent across the Agent Trust System.

4.2 Persistence and Non-Transferability

Agent Identities are persistent and non-transferable. An Agent Identity may not be reassigned to a different AI system. Where the underlying AI system changes materially — including, without limitation, a change of underlying model family, a material change of purpose, or a change of deploying organization — the Principal shall register a new Agent and shall not continue to operate under the prior Agent Identity.

4.3 Identity Record Contents

An Agent Identity record includes, at a minimum, the Agent identifier, the Principal identifier, the registration timestamp, a hash of the Agent Metadata, the Regulatory Classification, and the Trust Score (as periodically updated). Identity records committed on-chain are immutable as described in Section 12.

4.4 Credential Security

Principals are solely responsible for the security of credentials used to authenticate Agent API calls, including any API Keys, Access Tokens, or signing keys associated with the Agent. Compromise of such credentials shall be treated, and notified to Zoryxon, in accordance with the API Terms.

5. Authorization Framework

5.1 Authorization Grants

Principals delegate authority to Agents through Authorization Grants. Each Authorization Grant must specify, at a minimum:

• 5.1.1 the permitted actions (Scope);
• 5.1.2 the resource boundaries (including any applicable spending limits, data classes, or counterparties);
• 5.1.3 the time limits (start time and expiration);
• 5.1.4 the conditions, if any, under which the Authorization Grant is automatically revoked (including thresholds, anomaly conditions, or Compliance Check outcomes); and
• 5.1.5 any geographic or jurisdictional restrictions applicable to the Agent's actions under the Authorization Grant.

5.2 Least Privilege

Authorization Grants shall conform to the principle of least privilege. Principals shall grant only the minimum permissions necessary for the Agent's intended function and shall not configure Authorization Grants in a manner designed to circumvent future restrictions or to overstate required Scope.

5.3 Modification and Revocation

Principals (and, where so configured, designated Guardians) may modify or revoke Authorization Grants at any time. Revocation takes effect upon confirmation on the applicable blockchain network. Principals shall not rely on instantaneous off-chain effect for high-risk operations; where confirmation latency is material, Principals shall implement compensating controls.

5.4 Operating Within Scope

Agents shall not operate beyond the Scope of their Authorization Grants. Any action taken by an Agent outside the Scope of an Authorization Grant is the sole responsibility of the Principal, including any and all consequences thereof.

5.5 Boundary Enforcement

Zoryxon enforces authorization boundaries at the Zoryxon API and supported smart-contract layers. Zoryxon cannot guarantee enforcement of Authorization Grants in systems, networks, or counterparties outside the Zoryxon platform. Principals are responsible for designing Agents and integrations to respect Authorization Grants in all environments in which the Agent operates.

6. Compliance Audit Logging

6.1 Logging of Agent Actions

Agent actions processed through the Zoryxon platform are written to the Compliance Audit Log via the Hedera Consensus Service or any successor consensus layer designated by Zoryxon. Audit Records may be batched for cost efficiency without compromising the integrity, ordering, or timestamping of individual entries.

6.2 Audit Record Contents

Audit Records include, at a minimum, the Agent identifier, the action type, the timestamp, the Authorization Grant reference, the resource accessed, the outcome (including success or failure), and any relevant metadata necessary to evidence the action.

6.3 Immutability

Audit Records are immutable once written. They cannot be modified, deleted, or redacted by the Principal, the Guardian, or Zoryxon. The Principal acknowledges and accepts this immutability as a condition of enabling Compliance Audit Logging.

6.4 Retention

Audit Records committed to the Hedera Consensus Service are retained indefinitely on that network. Off-chain audit data, including enriched metadata and dashboard analytics, is retained in accordance with the data retention schedule set forth in the Privacy Policy and, where applicable, the Data Processing Agreement.

6.5 Export and Reporting

Zoryxon makes available export functionality permitting Principals to extract Audit Records for the Principal's own regulatory reporting and recordkeeping purposes. Zoryxon does not file regulatory reports on behalf of Principals and makes no representation that any export format is sufficient for any particular regulator's requirements.

6.6 Access Controls

Access to the Compliance Audit Log is restricted to the Principal, designated Guardians authorized by the Principal, and Zoryxon (acting for purposes of platform integrity, security, fraud prevention, and legal compliance). Principals are responsible for configuring appropriate access controls within their own organization.

7. Regulatory Classification

7.1 Categorization Framework

The Classification Engine categorizes Agents according to risk tiers aligned with the framework of the EU AI Act: Unacceptable Risk, High Risk, Limited Risk, and Minimal Risk. Classifications are produced based on Agent Metadata supplied by the Principal and the configured rules of the Classification Engine.

7.2 Informational and Advisory

Regulatory Classifications are informational and advisory only. They do NOT constitute a legal determination of any Agent's regulatory status under the law of any jurisdiction. The Classification Engine is a tool to assist, and is not a substitute for, qualified legal analysis.

7.3 Principal Responsibility

Principals are solely responsible for determining the actual regulatory classification of their Agents under applicable law, for fulfilling all applicable obligations associated with that classification, and for retaining qualified legal counsel where appropriate.

7.4 Unacceptable-Risk Agents

Agents that the Classification Engine identifies as, or that Zoryxon otherwise reasonably determines to be, within the Unacceptable Risk category may be flagged, restricted, suspended, or deregistered by Zoryxon, in its sole discretion, without liability to the Principal.

7.5 Methodology Updates

Regulatory Classifications may be updated as Agent Metadata changes or as regulatory frameworks evolve. Zoryxon shall provide reasonable notice of material changes to the classification methodology via the developer documentation, the Agent Dashboard, or direct communication.

8. Trust Score

8.1 Computation

Trust Scores are computed based on factors that may include, without limitation, compliance history, authorization adherence, audit log completeness, metadata accuracy, response to Compliance Checks, and other inputs as described in the developer documentation.

8.2 Informational Nature

Trust Scores are informational metrics provided as a convenience. They do NOT constitute an endorsement, certification, or guarantee of an Agent's reliability, safety, performance, or regulatory compliance, and shall not be relied upon as such by any party.

8.3 Methodology Updates

Zoryxon may modify the Trust Score methodology at any time. Material methodology changes shall be documented in the developer documentation and may be reflected in retrospective recalculation of Trust Scores.

9. Guardian System

9.1 Designation

Principals may designate one or more Guardians. Guardians are entities authorized by the Principal to monitor Agent activity and to undertake Interventions under defined conditions established by the Principal.

9.2 Guardian Powers

Guardian powers are configured by the Principal and may include, without limitation: viewing Audit Records, suspending Agent operations, revoking or modifying Authorization Grants, and triggering Compliance Checks. Guardian powers shall not exceed the powers granted by the Principal in the Guardian configuration.

9.3 Logging of Interventions

All Guardian Interventions are recorded in the Compliance Audit Log in accordance with Section 6, including the Guardian identifier, action type, timestamp, and outcome.

9.4 Selection and Configuration

Principals are solely responsible for selecting appropriate Guardians, defining appropriate Intervention conditions, and ensuring that Guardians have, and continue to have, the authority, competence, and resources to perform their assigned role.

9.5 Zoryxon Not a Guardian

Zoryxon does not act as a Guardian for any Agent except as may be required to protect platform integrity, the rights of other Users, or to comply with applicable law or legal process. Any action taken by Zoryxon in such circumstances shall not create a duty of supervision or oversight beyond that action.

10. Principal Responsibilities

10.1 Scope of Responsibility

The Principal is responsible for, and accepts all risk associated with:

• 10.1.1 the Agent's compliance with all applicable laws and regulations, including the EU AI Act and any equivalent or analogous regimes in jurisdictions where the Agent operates;
• 10.1.2 the accuracy and currency of all Agent Metadata;
• 10.1.3 the appropriateness, scope, and configuration of all Authorization Grants;
• 10.1.4 monitoring the Agent's behavior and responding promptly to anomalies, incidents, or Compliance Check outcomes;
• 10.1.5 maintaining the security of Agent credentials and integration infrastructure; and
• 10.1.6 all actions taken by the Agent, whether within or outside the Zoryxon platform, including actions that exceed the Scope of the Authorization Grant.

10.2 Acknowledgments

The Principal acknowledges that:

• 10.2.1 the Agent Trust System is technical infrastructure and not a compliance program;
• 10.2.2 use of the Agent Trust System does not guarantee, and shall not be relied upon as evidencing, regulatory compliance; and
• 10.2.3 the Principal must independently ensure compliance with the EU AI Act, U.S. federal and state AI laws and executive orders, sectoral AI rules, and all other applicable laws.

10.3 Insurance

The Principal shall maintain insurance coverage appropriate to the nature, scope, and risk profile of the Agent's operations, including, where applicable, technology errors and omissions, cyber liability, and general commercial liability coverage.

11. Prohibited Conduct

Without limiting the prohibitions set forth in the Acceptable Use Policy, the Terms of Service, or the API Terms, the Principal shall not, and shall not permit any Agent or any third party to:

• 11.1 register Agents with false, misleading, or materially incomplete Agent Metadata;
• 11.2 configure Authorization Grants that materially exceed the Agent's actual intended Scope for the purpose of circumventing future restrictions or oversight;
• 11.3 manipulate, falsify, suppress, or attempt to alter Compliance Audit Logs or Audit Records;
• 11.4 use the Classification Engine to misrepresent an Agent's Risk Category or other regulatory attributes;
• 11.5 operate Agents in jurisdictions where their operation is prohibited by applicable law, or in violation of applicable export control or sanctions law;
• 11.6 use the Agent Trust System to facilitate unauthorized surveillance, prohibited discrimination, social scoring, real-time remote biometric identification in publicly accessible spaces (except as permitted by applicable law), or any other application that constitutes an Unacceptable Risk under the EU AI Act or an analogous prohibited practice under any other applicable law;
• 11.7 circumvent or attempt to circumvent authorization enforcement, kill-switch mechanisms, Compliance Checks, or other safety features of the Agent Trust System;
• 11.8 impersonate any other Principal, Agent, Guardian, or Zoryxon personnel, or misrepresent affiliation with any of the foregoing; or
• 11.9 use the Agent Trust System in any manner that violates the Acceptable Use Policy or any other Zoryxon policy.

12. Data and Privacy

12.1 Application of Privacy Framework

Agent Metadata, Authorization Grants, Audit Records, and other data Processed through the Agent Trust System may contain or reference personal data. Processing of such data is governed by the Privacy Policy and, where applicable, the Data Processing Agreement.

12.2 On-Chain Data

On-Chain Records (including Agent identifiers, metadata hashes, Authorization references, and Audit Records committed to the Hedera Consensus Service) are written to public or semi-public blockchain networks and cannot be modified or deleted by Zoryxon, the Principal, or any Guardian. The Principal shall ensure that no unnecessary personal data, special-category data, confidential business information, or other sensitive content is included in fields that will be recorded On-Chain. Cryptographic hashes are one-way derivations and do not, on their own, constitute personal data.

12.3 Off-Chain Data

Off-Chain Records, including enriched metadata, dashboard analytics, and Agent configuration data, may be modified, restricted, or deleted in accordance with the Privacy Policy, the Data Processing Agreement, and applicable data protection law.

12.4 Data Subject Rights

Where Agents Process personal data of identifiable Data Subjects, the Principal shall, as Controller, respond to Data Subject requests in accordance with applicable law. Zoryxon shall assist as set forth in the Data Processing Agreement, subject to the limitations imposed by On-Chain immutability.

13. Fees

13.1 Inclusion in Subscription

Access to the Agent Trust System is included in qualifying Zoryxon subscription tiers. Additional fees may apply for, without limitation, high-volume Audit Logging, premium classification or analytics features, and enterprise-grade service-level coverage. Applicable fees are disclosed in the pricing materials and the Agent Dashboard.

13.2 Audit Logging Costs

Audit Records committed via the Hedera Consensus Service incur per-message infrastructure costs that are reflected in the Principal's subscription billing. Zoryxon employs message batching and other cost-optimization techniques without compromising the integrity of individual Audit Records.

13.3 Currency and Network Fees

All Zoryxon-charged pricing is denominated in United States dollars (USD). Blockchain network fees, if any, are separate and disclosed in the developer documentation. Principals are not required to hold cryptocurrency to use the Agent Trust System; applicable network fees, where assumed by Zoryxon, are billed in USD.

14. Service Limitations and Disclaimers

14.1 As Is and As Available

THE AGENT TRUST SYSTEM IS PROVIDED ON AN "AS IS" AND "AS AVAILABLE" BASIS, WITHOUT WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, NON-INFRINGEMENT, OR UNINTERRUPTED OR ERROR-FREE OPERATION.

14.2 No Warranty of Specific Outcomes

Zoryxon does not warrant that:

• 14.2.1 the Agent Trust System will be uninterrupted, timely, secure, or error-free;
• 14.2.2 Audit Records will be written within any particular timeframe (which depends on blockchain network conditions outside Zoryxon's control);
• 14.2.3 the Classification Engine will correctly classify any Agent under any jurisdiction's laws;
• 14.2.4 Trust Scores will accurately predict Agent behavior, performance, or compliance; or
• 14.2.5 the Agent Trust System will detect, prevent, or remediate all unauthorized, anomalous, or non-compliant Agent actions.

14.3 What the System Is Not

The Agent Trust System is NOT, and shall not be represented as: (a) a legal compliance program; (b) a regulatory certification body; (c) a substitute for qualified legal counsel; (d) an insurance product or guarantee against loss; or (e) a warranty of agent safety, reliability, or fitness for a particular purpose. Zoryxon is a technology infrastructure provider; regulatory compliance is and remains the sole responsibility of the Principal.

14.4 Mandatory-Law Carve-Out

Some jurisdictions do not allow the exclusion of certain warranties. In such jurisdictions, the foregoing exclusions apply to the maximum extent permitted by applicable law.

15. Indemnification

The Principal shall defend, indemnify, and hold harmless Zoryxon, its affiliates, and their respective officers, directors, employees, agents, and representatives (collectively, the "Zoryxon Indemnitees") from and against any and all claims, demands, losses, liabilities, damages, fines, penalties, costs, and expenses (including reasonable attorneys' fees) arising out of or relating to:

• 15.1 the Principal's Agents and any actions taken by such Agents, whether within or outside the Zoryxon platform and whether within or outside the Scope of any Authorization Grant;
• 15.2 the Principal's breach of these Agent Trust Terms or any other Zoryxon policy;
• 15.3 any regulatory inquiry, enforcement action, penalty, or fine related to the Principal's Agents, including under the EU AI Act or any analogous law;
• 15.4 third-party claims arising from the Principal's Agents' operations, including claims by Data Subjects, counterparties, or persons affected by Agent decisions;
• 15.5 the Principal's failure to maintain accurate Agent Metadata or appropriate Authorization Grants; and
• 15.6 the Principal's violation of any applicable law, including data protection, consumer protection, intellectual property, sanctions, and export control law.

16. Limitation of Liability

16.1 Aggregate Cap

TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, ZORYXON'S AGGREGATE LIABILITY ARISING OUT OF OR RELATING TO THESE AGENT TRUST TERMS, THE AGENT TRUST SYSTEM, OR THE PRINCIPAL'S USE OR INABILITY TO USE THE AGENT TRUST SYSTEM, WHETHER IN CONTRACT, TORT (INCLUDING NEGLIGENCE), STRICT LIABILITY, OR UNDER ANY OTHER THEORY, SHALL NOT EXCEED THE GREATER OF (A) THE FEES ACTUALLY PAID BY THE PRINCIPAL TO ZORYXON FOR AGENT TRUST SYSTEM ACCESS IN THE TWELVE (12) MONTHS PRECEDING THE EVENT GIVING RISE TO THE CLAIM, OR (B) ONE HUNDRED U.S. DOLLARS (US$100).

16.2 Excluded Damages

IN NO EVENT SHALL ZORYXON BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL, EXEMPLARY, OR PUNITIVE DAMAGES, OR FOR ANY LOSS OF PROFITS, REVENUE, GOODWILL, BUSINESS OPPORTUNITIES, USE, OR DATA, OR FOR THE COST OF PROCURING SUBSTITUTE SERVICES, ARISING OUT OF OR RELATING TO THESE AGENT TRUST TERMS OR THE AGENT TRUST SYSTEM, EVEN IF ZORYXON HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES.

16.3 Specific Exclusions

Without limiting Sections 16.1 or 16.2, Zoryxon shall not be liable for:

• 16.3.1 any regulatory penalties, fines, or sanctions imposed on the Principal by any governmental authority;
• 16.3.2 any damages arising from Agent actions, whether within or outside the Scope of an Authorization Grant;
• 16.3.3 any losses arising from Regulatory Classification outcomes, Trust Score computations, or Compliance Check results;
• 16.3.4 any damages arising from delays or failures of blockchain networks, including delays in writing Audit Records to the Hedera Consensus Service;
• 16.3.5 third-party payment processor, RPC provider, hosting provider, or other sub-processor failures; and
• 16.3.6 the Principal's failure to comply with these Agent Trust Terms or applicable law.

16.4 Mandatory-Law Carve-Out

Nothing in this Section 16 limits any liability that cannot be limited or excluded under applicable law, including liability for fraud or willful misconduct.

17. Suspension and Deregistration

17.1 Grounds

Zoryxon may, in its reasonable discretion, suspend or deregister any Agent that:

• 17.1.1 violates these Agent Trust Terms, the Acceptable Use Policy, or any other Zoryxon policy;
• 17.1.2 poses a security risk to the platform, its Users, or third parties;
• 17.1.3 is involved in, or facilitates, illegal activity;
• 17.1.4 has materially inaccurate, misleading, or outdated Agent Metadata; or
• 17.1.5 is the subject of valid legal process or a regulatory directive.

17.2 Notice

Zoryxon shall provide notice of suspension or deregistration where reasonably practicable, except where immediate action is necessary to address security threats, ongoing harm, fraud, or to comply with legal process or a regulatory directive.

17.3 Effect

Suspension or deregistration disables the Agent's access to the Zoryxon platform. Deregistration does not, and cannot, delete On-Chain Records associated with the Agent, including the Agent Identity record and the Compliance Audit Log. Off-Chain Records shall be handled in accordance with the Privacy Policy, the Data Processing Agreement, and applicable law.

18. Term and Termination

18.1 Term

These Agent Trust Terms are effective as of the date the Principal first registers an Agent and shall continue for so long as the Principal has one or more Agents registered in the Agent Trust System or otherwise accesses the Agent Trust System.

18.2 Deregistration by Principal

The Principal may deregister Agents at any time through the Agent Dashboard. Deregistration does not delete On-Chain Records associated with the deregistered Agent.

18.3 Termination by Zoryxon

Zoryxon may terminate the Principal's access to the Agent Trust System in accordance with the Terms of Service and Section 17.

18.4 Effect of Termination

Upon termination of these Agent Trust Terms or loss of access to the Agent Trust System:

• 18.4.1 all API access associated with affected Agents is revoked;
• 18.4.2 all active Authorization Grants associated with affected Agents are suspended or revoked;
• 18.4.3 On-Chain Records persist permanently in accordance with Sections 6 and 12;
• 18.4.4 Off-Chain Records shall be deleted within thirty (30) days of termination, except where retention is required by applicable law or by Zoryxon's ongoing legal obligations; and
• 18.4.5 the Principal remains responsible for actions taken by the Principal's Agents prior to termination.

18.5 Survival

The following Sections survive termination of these Agent Trust Terms: 2 (Definitions), 4.2 and 4.3 (with respect to On-Chain Identity Records), 5.4 (Operating Within Scope, with respect to actions taken prior to termination), 6 (Compliance Audit Logging, with respect to records already written), 10 (Principal Responsibilities, with respect to obligations arising prior to termination), 12 (Data and Privacy), 14 (Service Limitations and Disclaimers), 15 (Indemnification), 16 (Limitation of Liability), 17 (Suspension and Deregistration, with respect to its consequences), 18.4 and 18.5 (Effect and Survival), 20 (Governing Law and Dispute Resolution), 21 (Severability), and 22 (Contact).

19. Modifications

Zoryxon may modify these Agent Trust Terms from time to time. Zoryxon shall provide notice of material changes not less than thirty (30) days in advance of the effective date by means of in-product notice, email to the address on file, update of the "Last Updated" date with conspicuous notice in the Agent Dashboard, or any combination of the foregoing. Material changes may require the Principal to acknowledge the modified Agent Trust Terms before continuing to register or operate Agents. Continued registration of Agents or use of the Agent Trust System on or after the effective date constitutes acceptance of the modified Agent Trust Terms.

20. Governing Law and Dispute Resolution

20.1 Governing Law

These Agent Trust Terms are governed by, and shall be construed in accordance with, the laws of the State of Ohio, United States, without regard to its conflict-of-laws principles, except to the extent that federal law (including the Federal Arbitration Act) applies.

20.2 Binding Arbitration

Any dispute arising out of or relating to these Agent Trust Terms shall be resolved by binding arbitration in accordance with the dispute resolution provisions of the Terms of Service. Arbitration shall be administered by the American Arbitration Association under its Commercial Arbitration Rules, shall be seated in Canton, Ohio (Stark County), and shall be conducted by a single arbitrator. Judgment upon the award rendered by the arbitrator may be entered in any court of competent jurisdiction.

20.3 Class-Action Waiver

EACH PARTY AGREES THAT ANY DISPUTE RESOLUTION PROCEEDINGS WILL BE CONDUCTED ONLY ON AN INDIVIDUAL BASIS AND NOT AS A CLASS, CONSOLIDATED, OR REPRESENTATIVE ACTION. The arbitrator may not consolidate more than one person's claims and may not preside over any form of representative or class proceeding.

20.4 Carve-Outs

Nothing in this Section 20 prevents either party from (a) seeking injunctive or other equitable relief in any court of competent jurisdiction with respect to intellectual property rights or unauthorized access, or (b) bringing individual claims in small-claims court that qualify for adjudication therein.

21. Severability

If any provision of these Agent Trust Terms is held by a court or arbitrator of competent jurisdiction to be invalid, illegal, or unenforceable in any respect, that provision shall be enforced to the maximum extent permissible so as to effect the intent of the parties, and the validity, legality, and enforceability of the remaining provisions shall not in any way be affected or impaired.

22. Contact

All notices and inquiries relating to these Agent Trust Terms should be addressed to: