AI Usage Policy

IMPORTANT: AI features can produce inaccurate, incomplete, biased, or unsuitable outputs. Customers must apply human review and independent judgment before relying on or publishing AI-generated content. Current versions of HELIX customer agreements, policies, and compliance documents are available through the HELIX Legal & Compliance Center at https://gohelix.ai/legal.

1. PURPOSE AND SCOPE

This AI Usage Policy (the “Policy”) governs access to and use of artificial intelligence, machine learning, generative AI, automated decisioning, predictive, classification, transcription, summarization, recommendation, and similar features made available through HELIX (collectively, “AI Features”). It applies to Customers, Authorized Users, end users, contractors, and any person using AI Features through a HELIX account.

This Policy supplements and is incorporated into the applicable HELIX Master Services Agreement, HELIX Platform Terms of Service, HELIX Acceptable Use Policy, HELIX Privacy Policy, HELIX Data Processing Addendum, and HELIX Information Security Policy. Capitalized terms not defined here have the meanings assigned in those documents.

2. KEY PRINCIPLES

Human accountability. AI may assist people, but it does not replace responsible human judgment, review, approval, or legal accountability.

Risk-based use. The level of oversight, testing, documentation, and approval must increase as the potential impact on individuals, customers, or the business increases.

Privacy and security by design. Users must minimize data shared with AI Features, use approved tools, protect credentials, and follow applicable privacy and security requirements.

Transparency. Users must not intentionally misrepresent material AI-generated content as verified human analysis when disclosure is legally required or reasonably necessary to avoid deception.

Fairness and non-discrimination. AI Features must not be used to unlawfully discriminate, profile, manipulate, or cause unfair treatment.

Accuracy and validation. Outputs must be reviewed for factual accuracy, context, intellectual property concerns, and fitness for the intended purpose.

3. CUSTOMER RESPONSIBILITY AND HUMAN OVERSIGHT

Customer remains solely responsible for its use of AI Features, all prompts and inputs, all outputs, and all actions taken based on those outputs. Customer must establish human review appropriate to the use case and may not rely on AI output as the sole basis for a decision that could materially affect a person’s legal rights, access to employment, housing, credit, insurance, healthcare, education, essential services, or other significant interests unless expressly authorized in writing and supported by legally sufficient safeguards.

Customer must verify AI output before it is sent, published, implemented, used in a campaign, incorporated into a deliverable, or relied upon for a business decision. Review must include, as applicable:

factual accuracy and completeness;

tone, context, and brand suitability;

legal, regulatory, professional, and industry requirements;

privacy, confidentiality, consent, and data-use restrictions;

bias, discrimination, and accessibility concerns;

copyright, trademark, publicity, and other third-party rights;

security implications, including unsafe code, links, instructions, or automations; and

appropriate disclosure that AI was used, when required by law, contract, professional standard, or context.

4. PERMITTED USES

Subject to this Policy and the other governing documents, AI Features may be used for legitimate business purposes such as:

drafting and revising marketing, sales, customer-service, and operational content;

summarizing information and communications;

classification, tagging, routing, and workflow assistance;

brainstorming, ideation, outlining, and research support;

transcription and meeting-note assistance where legally permitted and appropriately disclosed;

data analysis, trend identification, and forecasting subject to validation;

assisting with software, website, workflow, and automation development; and

other approved uses consistent with the Customer’s business, legal obligations, and configured HELIX services.

5. PROHIBITED AND RESTRICTED USES

Customer and Authorized Users must not use AI Features to engage in any activity prohibited by the HELIX Acceptable Use Policy or applicable law. Without limiting those restrictions, AI Features may not be used to:

create, facilitate, or conceal unlawful, fraudulent, deceptive, defamatory, abusive, exploitative, or malicious activity;

impersonate a real person or falsely present fabricated statements, recordings, images, or communications as authentic;

generate or distribute malware, credential theft, phishing, evasion techniques, unauthorized surveillance, or instructions designed to compromise systems;

infer, rank, or make decisions based on protected characteristics or highly sensitive traits in a manner that is unlawful, discriminatory, or unfair;

conduct high-impact automated decisioning without legally required notice, testing, human review, appeal mechanisms, and other safeguards;

produce or facilitate sexual exploitation, child sexual abuse material, non-consensual intimate content, or content that sexualizes minors;

generate targeted political persuasion or voter suppression activity in violation of law or provider rules;

provide individualized legal, medical, financial, tax, employment, or other professional advice as a substitute for a qualified professional;

create false reviews, fabricated testimonials, fake endorsements, undisclosed synthetic spokespersons, or deceptive engagement;

upload or process data that Customer is not authorized to use, including confidential third-party information, trade secrets, protected health information, financial account credentials, government identifiers, authentication secrets, or other restricted data unless the applicable service and written agreement expressly authorize it;

bypass safety controls, usage limits, access controls, provider policies, or technical restrictions; or

use AI output to train, develop, or improve a competing platform or model using HELIX Work Product, confidential information, or restricted third-party content.

6. HIGH-IMPACT AND REGULATED USE CASES

AI use involving employment, credit, lending, insurance, healthcare, housing, education, public benefits, biometric identification, legal services, safety-critical systems, or other regulated or high-impact decisions requires advance written approval from HAVOK and may require a separate agreement, specialized configuration, impact assessment, legal review, testing, monitoring, and customer-facing notices.

HAVOK may prohibit, restrict, suspend, or condition any AI use that creates unacceptable legal, security, ethical, operational, or third-party provider risk. Availability of an AI Feature does not mean that the Feature is approved for every industry, dataset, jurisdiction, or use case.

7. DATA, PRIVACY, AND CONFIDENTIALITY

Customer must use data minimization and submit only the information reasonably necessary for the intended task. Customer is responsible for having a lawful basis, providing required notices, obtaining required consents, honoring applicable rights, and complying with contractual restrictions before submitting personal data or confidential information to an AI Feature.

Customer must not include passwords, API keys, payment-card data, bank credentials, private encryption keys, authentication tokens, or other secrets in prompts or inputs. Sensitive or regulated data may be submitted only where the relevant service configuration, third-party terms, security controls, and written agreements expressly permit that processing.

AI Features may rely on third-party AI providers. Data handling may vary by provider, feature, account configuration, region, and service tier. The HELIX Privacy Policy and HELIX Data Processing Addendum describe HELIX’s general processing practices, and third-party provider terms may also apply.

8. AI PROVIDERS AND THIRD-PARTY TECHNOLOGY

HELIX may integrate AI technology, models, APIs, hosting, or supporting services supplied by third parties. HAVOK does not control all aspects of third-party models or services and does not warrant their continued availability, training data, model behavior, accuracy, security, non-infringement, or regulatory status.

HAVOK may add, replace, modify, limit, or discontinue an AI provider or AI Feature to address security, cost, compliance, performance, availability, or product considerations. Customer must comply with applicable third-party terms and provider policies.

9. OUTPUT QUALITY, LIMITATIONS, AND NO PROFESSIONAL ADVICE

AI output is probabilistic and may be inaccurate, incomplete, outdated, biased, offensive, inconsistent, or unsuitable. AI Features may fabricate facts, citations, sources, calculations, code, or claims. Similar or identical outputs may be generated for other users.

AI Features are provided as tools to assist users and do not provide legal, medical, financial, tax, employment, regulatory, security, or other professional advice. Customer must obtain advice from qualified professionals where appropriate and must independently validate material claims, citations, calculations, and recommendations.

10. INTELLECTUAL PROPERTY AND CONTENT RIGHTS

Customer retains its rights in Customer Data and is responsible for ensuring it has the rights necessary to submit all inputs. HAVOK and its licensors retain their rights in the Platform, AI Features, models, methods, prompts, templates, workflows, configurations, documentation, and HELIX Work Product.

To the extent permitted by the applicable provider terms and law, Customer may use AI output generated for Customer through the Services. HAVOK does not represent or warrant that output is unique, protectable by intellectual property law, or free from third-party claims. Customer is responsible for reviewing output for potential infringement, attribution, licensing, confidentiality, or publicity-right concerns before use.

Customer may not use AI Features to remove copyright management information, misappropriate protected content, create unauthorized derivative works, or intentionally imitate a living artist, identifiable person, brand, or proprietary style in a manner that violates law or third-party rights.

11. CODE, AUTOMATIONS, AND ACTION-TAKING AI

AI-generated code, scripts, workflows, configurations, and automation logic must be tested in a safe environment before production use. Customer must review permissions, data access, security, error handling, rollback procedures, message recipients, timing, and potential unintended actions.

Customer must use heightened safeguards for AI Features that can send communications, modify records, initiate payments, publish content, access external systems, create accounts, or take other actions. Appropriate controls may include approval gates, role-based access, transaction limits, test environments, monitoring, logging, and emergency disablement.

12. DISCLOSURE, RECORDKEEPING, AND AUDITABILITY

Customer must maintain records reasonably necessary to demonstrate compliance for material or high-risk AI use, including the business purpose, responsible owner, data categories, testing performed, human-review process, known limitations, approvals, and incident history.

Where legally required or necessary to avoid deception, Customer must clearly disclose the use of AI-generated or materially AI-altered content. Disclosures must be understandable, appropriately placed, and not contradicted by other statements or design elements.

13. SECURITY AND AI-SPECIFIC RISKS

Customer must protect AI Features against prompt injection, data exfiltration, unauthorized tool use, malicious files, unsafe links, insecure code, model manipulation, and other AI-specific threats. Users should treat untrusted content as potentially adversarial and must not allow AI Features to access systems or data beyond what is necessary.

Customer must promptly report suspected misuse, unauthorized disclosure, harmful output, security incidents, or material AI failures to HAVOK using the contact information below. HAVOK may investigate, restrict features, preserve logs, or take protective action consistent with the governing agreements and applicable law.

14. MONITORING, TESTING, AND CONTINUOUS IMPROVEMENT

HAVOK may evaluate AI Features for performance, safety, security, abuse prevention, and product improvement using methods permitted by the governing agreements and applicable law. Testing may include automated and human review, red-team exercises, quality sampling, incident analysis, and provider monitoring.

HAVOK may update controls, models, prompts, filters, limits, interfaces, and policies as technology, risks, laws, and provider requirements evolve. No AI system can eliminate all risk, and safeguards may produce false positives, false negatives, or unexpected behavior.

15. ENFORCEMENT

HAVOK may investigate suspected violations and may remove content, disable an AI Feature, restrict capabilities, suspend access, require remediation, or terminate Services when reasonably necessary to protect users, third parties, providers, HAVOK, or the Platform. Enforcement decisions may consider severity, repetition, intent, impact, cooperation, and remediation.

Customer remains responsible for Fees and other obligations during a suspension, as provided in the governing agreements. Nothing in this Policy limits HAVOK’s rights or remedies under the HELIX Master Services Agreement, HELIX Platform Terms of Service, HELIX Acceptable Use Policy, law, or third-party provider requirements.

16. POLICY UPDATES

HAVOK may update this Policy to reflect changes in AI technology, law, regulation, provider requirements, security practices, or the Services. Material updates will be communicated or posted as required by the governing agreements or applicable law. Continued use of AI Features after clear notice of an effective update constitutes acceptance where legally permitted. Payment of an invoice acknowledges the applicable billing notice but does not replace affirmative acceptance procedures where such procedures are required.

17. ORDER OF PRECEDENCE

If this Policy conflicts with a signed agreement, the signed agreement controls. The HELIX Data Processing Addendum controls solely with respect to the processing of personal data within its scope. The HELIX Acceptable Use Policy controls as to broader prohibited conduct. This Policy controls as to AI-specific use requirements unless a more specific written agreement states otherwise.

18. CONTACT AND REPORTING

Questions about this Policy or reports of suspected AI misuse, unsafe output, security issues, or unauthorized data disclosure should be sent to: