AI Use Policy

This AI Use Policy explains the rules that apply when customers, developers, End Users, contractors, and other users access or use Lythe products and services provided by Lythe Inc. or its applicable affiliate, including Lythe Pte. Ltd. (collectively and as applicable, "Lythe"). This includes Fraser Simulation Studio, AI personas, subagents, simulations, automated evaluation tools, bug-testing workflows, visual evaluations, Simulation Evaluation tools, APIs, integrations, dashboards, transcripts, analytics, public demos, and related services (collectively, the "Services"). The Services may also process call recordings, transcripts, and conversation metadata provided by customers for AI persona generation and simulation, but the Services do not provide outbound calling, telephony services, voice agents, or message delivery.

For use of Lythe's self-serve offerings, Lythe Inc. is the entity providing the Services worldwide. Lythe Pte. Ltd. may provide Services only where it is separately and expressly identified as the responsible entity in an applicable Order Form, customer agreement, Terms of Service, or other written agreement.

The purpose of this policy is to help ensure that the Services are used safely, lawfully, transparently, and reliably. Users may not use the Services to deceive people; misrepresent synthetic feedback as real user research; violate applicable privacy, data protection, consumer protection, intellectual property, or AI disclosure laws; deceptively impersonate any person, business, government entity, institution, customer, expert, or protected group; manipulate vulnerable individuals; or deploy AI personas, subagents, simulations, or automated evaluators without appropriate oversight, disclosure, and controls; or test, crawl, scrape, or monitor third-party websites, applications, systems, prototypes, or APK files without authorization.

This policy applies to all access to and use of the Services, whether through the web application, APIs, SDKs, integrations, simulation tools, developer portals, dashboards, public demos, internal tools, beta features, customer pilots, enterprise trials, or production deployments.

Customers are responsible for their users, agents, AI personas, subagents, simulations, prompts, workflows, scripts, knowledge bases, integrations, leads, Customer Data, uploaded materials, and downstream use of outputs, including ensuring that their use of the Services complies with applicable laws, regulations, contractual obligations, and this policy.

Developers are responsible for how the Services are integrated into or made available through their products, workflows, and customer experiences, including implementing appropriate disclosures, permissions, safeguards, monitoring, and human oversight.

These responsibilities apply in addition to, and do not limit, Lythe's own obligations under applicable law.

Lythe may investigate, suspend, restrict, or terminate access to the Services where use violates this policy or creates legal, safety, security, compliance, reputational, or platform-integrity risk.

For purposes of this AI Use Policy, the following terms have the meanings set out below:

"AI Agent" means an automated or semi-automated system that receives inputs, reasons over context, generates outputs, takes actions, or interacts with individuals or systems through text, APIs, software tools, or other interfaces.

"AI Persona" means an AI-generated, AI-assisted, or customer-configured simulated user, customer profile, evaluator, reviewer, tester, respondent, or behavioral model used to test, review, critique, or analyze Customer Data, products, websites, user flows, prototypes, positioning, messaging, customer journeys, or related materials.

"Simulation Evaluation" means the use of multiple agents, AI personas, subagents, simulations, tests, red-teamers, automated evaluators, reviewers, or similar evaluation systems to test, compare, monitor, improve, or stress-test the behavior, performance, reliability, or safety of AI personas, subagents, simulations, automated evaluators, product experiences, user flows, messaging, positioning, customer journeys, or related workflows.

"Customer Data" means any data, information, content, or materials submitted to, processed by, generated through, stored by, or retrieved by or through the Services on behalf of a customer, including prompts, scripts, call recordings, transcripts, CRM records, lead information, product flows, websites, screenshots, prototypes, landing pages, product concepts, research notes, customer feedback, user personas, files, URLs, logs, metadata, analytics, uploaded materials, and integration data.

"End User" means any person who interacts with an AI persona, simulation, public demo, or other Lythe service, including leads, customers, patients, constituents, applicants, employees, research subjects, users, or other individuals.

"Output" means responses, transcripts, classifications, recommendations, analytics, evaluations, scores, simulated feedback, product critiques, insights, reports, actions, or other content generated by the Services, AI personas, or simulations.

"High-Impact Decision" means any decision, recommendation, assessment, ranking, determination, or action that may materially affect an individual's employment, credit, housing, healthcare, insurance, education, legal status, financial access, public benefits, safety, or access to similar rights, services, or opportunities.

"Regulated Use Case" means any use of the Services involving legal, medical, financial, insurance, housing, employment, education, government, political, children or minors, emergency, or other activities or contexts subject to heightened legal, regulatory, professional, or safety requirements.

4.1 Transparency: Customers must ensure that End Users understand when they are interacting with an AI system where disclosure is required by applicable law, regulation, platform policy, customer agreement, Lythe policy, or where non-disclosure would be materially deceptive.

4.2 Lawful Consent: Customers must obtain and maintain legally sufficient consent, permission, authorization, or other lawful basis for collecting, uploading, and processing call recordings, transcripts, communications data, uploaded materials, simulations, and AI processing where required by applicable law or regulation.

4.3 Human Accountability: Customers remain responsible for the AI personas, subagents, simulations, automated evaluators, prompts, workflows, and outputs they deploy or rely on.

4.4 Safety by Design: Customers must appropriately scope, test, monitor, log, and supervise AI personas, subagents, simulations, and automated evaluators and must implement human escalation where appropriate for the intended use case.

4.5 No Deception: The Services must not be used to mislead people about identity, authority, capability, intent, sponsorship, research methodology, source of feedback, or material facts.

4.6 Data Minimization: Customers must limit the collection and processing of Customer Data to what is necessary for the stated workflow and must avoid unnecessary processing of sensitive personal data.

Subject to applicable laws, customer agreements, and this policy, the Services may be used for legitimate business workflows such as:

• Summarizing and analyzing customer-provided call recordings, transcripts, and conversation metadata (with necessary consent) for internal research, persona generation, simulation, and product evaluation.
• Internal business operations such as research preparation, QA reviews, workflow automation, and CRM updating where proper consent exists.
• Customer support, frequently asked questions, ticket creation, and status updates.
• Business analytics, process improvement, and other internal operations.
• Product feedback simulations, landing page reviews, user flow testing, onboarding critiques, positioning analysis, messaging analysis, feature evaluation, product-market-fit hypotheses, bug testing, and visual evaluations of websites, applications, and product flows.
• Agent testing, simulation, red-teaming, prompt evaluation, performance improvement, and reliability testing through Simulation Evaluation.
• AI simulation pilots in regulated industries, provided that customers implement appropriate compliance controls, obtain any required internal approvals, maintain appropriate human oversight, and obtain any prior written approval required from Lythe under Section 14.

Users may not use Lythe services, including Fraser Simulation Studio, for any of the following activities:

6.1 Illegal, Harmful, or Abusive Activity

• Violating any applicable law, regulation, court order, industry rule, platform policy, or contractual obligation.
• Fraud, scams, phishing, social engineering, identity theft, credential theft, account takeover, or deceptive collection of personal data.
• Harassment, threats, intimidation, extortion, hate, abuse, stalking, doxxing, or non-consensual surveillance.
• Generating or facilitating malware, unauthorized access, evasion, spam, denial-of-service attacks, or other security abuse.
• Simulations, outputs, or automated workflows that are unlawful, unsolicited, misleading, coercive, deceptive, or abusive.

6.2 Deceptive AI, Human Impersonation, or Synthetic Research Misrepresentation

• Deploying an AI persona, subagent, simulation, or automated evaluator that falsely claims to be a specific human, licensed professional, government official, emergency responder, employee, customer, expert, public figure, or representative without authorization.
• Designing an AI persona, subagent, simulation, or automated evaluator to deny, hide, or misrepresent that it is AI when disclosure is legally required or when the context would make non-disclosure materially deceptive.
• Using cloned, synthetic, or similar-sounding voices without valid authorization from the relevant person or rights holder.
• Misrepresenting identity, business affiliation, legal authority, research methodology, or the purpose of the interaction.
• Creating, presenting, or distributing simulated feedback, AI persona responses, or outputs as real user feedback, real customer interviews, verified market research, or statistically valid survey results without appropriate disclosure and validation.

6.3 Telemarketing, Messaging, and Communications Abuse

• Using the Services to place outbound calls or messages or to engage in telemarketing, robocalling, automated outreach, or messaging campaigns, or to facilitate or enable such communications in violation of applicable law or without legally sufficient consent. The Services do not provide telephony or messaging capabilities.
• Attempting to bypass or ignore opt-out, suppression, user privacy choices, or data subject rights requirements.
• Using the Services to conduct spam campaigns, mass unsolicited outreach, marketing campaigns, or abusive sales or debt collection tactics.
• Circumventing platform safety controls, rate limits, monitoring, compliance systems, or other abuse-prevention measures.

6.4 High-Impact and Regulated Decisions

• Using Lythe services as the sole basis for any High-Impact Decision, or to make or materially influence a High-Impact Decision without appropriate human review, documentation, compliance controls, and any prior written approval required by Lythe under Section 14.
• Representing an AI persona, subagent, simulation, or automated evaluator as a licensed professional, or providing regulated medical, legal, financial, insurance, employment, housing, or public-benefits advice without required authorization, disclosures, compliance controls, and qualified human oversight.
• Making eligibility, diagnosis, treatment, credit, insurance, hiring, firing, tenant screening, immigration, or law-enforcement decisions without required compliance controls.
• Using Lythe services for emergency dispatch, crisis response, or situations where failure could create immediate risk to life or safety, unless expressly approved in writing by Lythe under a specific regulated deployment agreement.

6.5 Sensitive, Vulnerable, or Protected Groups

• Exploiting, manipulating, or unfairly targeting children, older adults, people in crisis, or other individuals in circumstances of vulnerability.
• Inferring or using sensitive characteristics such as race, religion, health, disability, sexuality, political opinion, union membership, or biometric identity in ways that violate law or create unfair treatment.
• Deploying AI personas, subagents, simulations, or automated evaluators to manipulate political views, suppress voting, mislead voters, impersonate campaigns or officials, or conduct unlawful political outreach.

6.6 Model, System, Platform, and Research Abuse

• Attempting to reverse engineer, bypass, overload, degrade, probe, scrape, or misuse Lythe systems or infrastructure.
• Using prompts, tool calls, data, or integrations to bypass safety filters, rate limits, access controls, monitoring, compliance controls, or billing systems.
• Using Simulation Evaluation or other automated improvement features to coordinate spam, evasion, manipulation, artificial engagement, security abuse, or platform attacks.
• Creating benchmark, evaluation, or comparison claims that are misleading, non-reproducible, or falsely attributed to Lythe services.
• Using public demos or Services to crawl, scrape, test, evaluate, overload, or analyze third-party websites, applications, prototypes, products, or systems without authorization.

Customers must comply with all applicable AI disclosure and consumer-protection requirements. Where required by law, regulation, platform policy, customer agreement, Lythe instruction, or where non-disclosure would be materially deceptive, customers must clearly disclose that an End User is interacting with an AI system or that feedback, evaluations, personas, or outputs are AI-generated or simulated.

Lythe recommends that AI personas or subagents identify themselves at the start of an interaction using clear and plain language appropriate to the use case. For example:

"Hello, I am an AI persona engaged to assist with product research on behalf of [Company]. This conversation may be recorded. You may request to speak with a human representative at any time."

For AI personas, simulations, and automated evaluations, customers must use clear and accurate language when describing outputs, assumptions, limitations, and research methodology. Customers must not imply that synthetic feedback is real user research, verified customer feedback, statistically valid survey data, or guaranteed market truth unless independently validated through appropriate research methods.

These examples are provided for illustration only. Customers remain responsible for determining and implementing any additional disclosures, consents, recording notices, opt-out mechanisms, research notices, or human-escalation statements required for their use case and jurisdiction.

Customers are responsible for determining whether and how consent, notice, registration, or other compliance requirements apply to collecting, uploading, and processing call recordings, transcripts, conversation metadata, communications, customer feedback, uploaded materials, and integrations.

• Customers must obtain, document, maintain, and honor all consents required under applicable law before collecting, uploading, or processing call recordings, transcripts, conversation data, or personal data through the Services.
• Customers must comply with applicable privacy and data-protection laws, user notice requirements, opt-out obligations, marketing rules, and regulatory requirements when using the Services.
• Customers must promptly honor opt-out or consent revocation requests and maintain suppression records where required.
• Customers must provide accurate disclosures, workflow descriptions, research methodology, business identity information, and compliance documentation to Lythe where requested.
• Lythe may require proof of consents, data-source documentation, workflow descriptions, research methodology, business identity, opt-out procedures, and compliance controls before enabling or continuing certain use cases.

Customers must design AI personas, subagents, simulations, automated evaluators, and workflows with appropriate human oversight. This includes clear escalation paths where an interaction or output involves uncertainty, user distress, complaints, legal threats, regulated advice, account changes, payment disputes, safety issues, research claims, or requests to speak with a human.

• AI personas, subagents, simulations, and automated evaluators must have defined authority limits and must not promise outcomes they cannot guarantee.
• AI personas, subagents, simulations, and automated evaluators must escalate to an appropriate human reviewer where confidence is low, where an End User objects to interacting with AI, or where a request falls outside the approved scope.
• Customers must regularly review outputs, transcripts, simulation results, failure modes, complaints, opt-outs, and escalations, subject to applicable law.
• Customers must maintain accountability and appropriate approval controls for workflows that trigger CRM updates, bookings, cancellations, payments, account changes, research conclusions, product decisions, or business decisions.

Customers are responsible for all prompts, scripts, knowledge bases, websites, files, FAQs, CRM records, customer records, research data, call recordings, transcripts, conversation metadata, datasets, product flows, screenshots, prototypes, landing pages, user research materials, APIs, and other content they upload, connect, or make available to Lythe services.

• Customer content must be accurate, lawful, authorized, and suitable for the intended workflow.
• Customers must not upload or make available personal data, confidential data, health data, financial data, children's data, or other sensitive data unless they are authorized to do so, such processing is necessary for the intended use case, and any approval required under Section 14 has been obtained.
• Customers must not use third-party content, voices, trademarks, recordings, datasets, websites, product materials, customer records, or personal data without appropriate rights.
• Customers must test AI persona, subagent, simulation, automated evaluator, and output behavior before launch and whenever prompts, tools, integrations, models, datasets, or knowledge sources materially change.

Lythe services may offer Simulation Evaluation, simulation, testing, optimization, monitoring, red-teaming, or automated improvement features. These features are intended to improve the reliability, safety, performance, usability, and compliance of AI personas, subagents, simulations, automated evaluators, product experiences, and related workflows.

Customers may not use Simulation Evaluation or other automated improvement features to evade compliance requirements; manipulate or deceive people; coordinate spam, harassment, synthetic research abuse, abusive communications, unauthorized testing activity, or security abuse; create harmful autonomous behavior; misrepresent synthetic feedback as real research; or deploy materially higher-risk behavior without appropriate controls and approval.

Customers must define appropriate success metrics, failure conditions, prohibited behaviors, escalation thresholds, human-review requirements, and rollback procedures before enabling automated optimization or improvement in production workflows, End User interactions, or externally relied-on research outputs.

Automated improvements must not be deployed in connection with High-Impact Decisions or Regulated Use Cases without appropriate testing, documented review, human approval, ongoing monitoring, and rollback controls.

Customers must ensure that any use of Customer Data, personal data, call recordings, transcripts, research data, product materials, or live End User interactions for evaluation, simulation, optimization, or automated improvement is lawful, authorized, appropriately secured, and consistent with applicable disclosures, consents, and contractual obligations.

Customers must maintain records reasonably necessary for auditability, including, where applicable, evaluation logs, test results, prompt and workflow versions, model or configuration changes, approval records, deployment decisions, identified failures, and remediation actions.

Lythe may investigate, restrict, pause, disable, or require review of Simulation Evaluation, automated optimization, or automated improvement features where Lythe reasonably determines that their use violates this policy or creates legal, safety, security, compliance, reliability, reputational, or platform-integrity risk.

Customers must use Lythe services in compliance with applicable privacy, data protection, and security laws. Customers are responsible for providing required notices, obtaining consents, honoring rights requests, configuring data retention, and limiting access to Customer Data.

• Do not submit data that you are not authorized to process.
• Do not use Lythe services to collect sensitive personal data through deception or without appropriate disclosure.
• Do not connect systems or APIs in a way that exposes data beyond what is necessary for the workflow.
• Use appropriate authentication, access controls, environment separation, and internal approval processes.
• Promptly notify Lythe of any suspected unauthorized access, misuse, security incident, privacy violation, or harmful agent behavior involving Lythe services or Customer Data.

Customers must test AI personas, subagents, simulations, and automated evaluators before production use and monitor them after deployment. Testing must appropriately cover accuracy, refusal behavior, escalation, consent language, regulatory disclosures, tool use, CRM updates, research methodology, edge cases, adversarial inputs, and failure recovery, taking into account the nature and risk of the intended use case.

• Customers must maintain records of production AI persona versions, subagent configurations, prompts, scripts, tools, knowledge bases, simulation settings, evaluation criteria, and deployment dates.
• Customers must conduct monitoring appropriate to the nature and risk of the deployment, which may include reviewing call recordings, transcripts, simulation outputs, reports, or logs where permitted by applicable law and customer policy.
• Customers must investigate abnormal outcomes, user complaints, opt-outs, hallucinations, research errors, unreliable outputs, or AI persona, subagent, simulation, or automated evaluator actions outside approved scope.
• Customers must pause, disable, or roll back AI personas, subagents, simulations, automated evaluators, or workflows where material legal, safety, security, compliance, or reliability risk is identified.

Customers must not deploy Lythe services for a Regulated Use Case without Lythe's prior written approval, unless Lythe expressly states in writing that prior approval is not required for a particular category of use.

Regulated Use Cases include, without limitation, healthcare, financial services, insurance, legal services, employment, housing, education, government services, political outreach, debt collection, emergency response, and workflows involving minors or individuals in circumstances of vulnerability.

Approval may require additional terms, security review, human oversight, audit logs, compliance documentation, business verification, customer prompts, scripts, workflow descriptions, opt-out procedures, research methodology, data retention controls, restrictions on automated decision-making, or other safeguards appropriate to the proposed use.

Lythe may investigate potential violations of this policy. Lythe may suspend, limit, throttle, block, disable, or terminate access to the Services, AI personas, subagents, simulations, automated evaluators, APIs, integrations, workflows, uploads, reports, or accounts where Lythe reasonably believes there is a violation, risk of harm, legal exposure, security threat, or platform abuse.

Lythe may also remove content, require remediation, request proof of consent, request proof of rights or data source, notify affected parties where required, cooperate with law enforcement or regulators where legally required, or preserve records for compliance and enforcement purposes.

To report suspected misuse of Lythe services, including unlawful uses, impersonation, spam, fraud, unsafe AI behavior, misleading synthetic research, privacy violations, unauthorized third-party website testing or analysis, or security concerns, contact: team@lythe.ai.

This AI Use Policy supplements the applicable Lythe Terms of Service, Terms of Use, customer agreements, Order Forms, and any applicable product-specific terms that expressly incorporate or reference this policy, including terms for Fraser Simulation Studio.

For use of Lythe's self-serve offerings, Lythe Inc. is the entity responsible for providing the Services worldwide. Lythe Pte. Ltd. may provide Services only where it is separately and expressly identified as the responsible entity in an applicable Order Form, customer agreement, Terms of Service, or other written agreement.

If this policy conflicts with a signed customer agreement, the signed customer agreement controls to the extent of the conflict. Notwithstanding the foregoing, the applicable Lythe entity may take actions reasonably necessary to comply with applicable law or to protect the safety, security, integrity, or lawful operation of Lythe services, subject to the applicable customer agreement and applicable law.

Nothing in this AI Use Policy limits or modifies the obligations of Lythe Inc. or Lythe Pte. Ltd., as applicable, under any applicable Privacy Policy, Data Processing Addendum, data processing agreement, or applicable data protection law.

Lythe may update this AI Use Policy from time to time to reflect product changes, legal developments, AI requirements, model behavior, safety learnings, or operational needs. Updated versions will be posted or otherwise made available to affected customers and users.

Unless otherwise required by applicable law or agreed in writing, material changes will apply prospectively from the effective date stated in the updated policy. Continued use of Lythe services after the effective date of an updated policy constitutes acceptance of the updated policy, subject to any contrary terms in a signed customer agreement.

Questions about this AI Use Policy may be sent to:

• Legal: team@lythe.ai
• Abuse reports: team@lythe.ai
• Privacy: team@lythe.ai
• Security: team@lythe.ai

The Lythe entity responsible for your use of the Services is identified in your applicable Terms of Service, Order Form, customer agreement, or other written agreement.