ZexirAI · Acceptable Use

Build boldly. Do not use it to harm.

This policy sets boundaries for lawful, safe, and responsible use of ZexirAI, including rules for child safety, violence, cyber activity, fraud, privacy, intellectual property, automated decisions, platform integrity, and regulated activity.

How we evaluate use: Context, intent, capability, scale, safeguards, and foreseeable impact all matter. Research, prevention, and fictional discussion can be allowed even when an operational request that meaningfully enables harm is not.

1. Principles and scope

This Acceptable Use Policy applies to every use of ZexirAI, including prompts, generated output, files, images, audio, code, project work, shared memory, search, connected services, automated workflows, and attempts to access the Service. It applies whether you use a Free or paid plan, an individual or organizational account, standard or Private mode, a browser or native application, or direct and automatically routed models.

You must comply with law, respect the rights and safety of others, and use safeguards proportionate to risk. You may not ask another person, model, tool, or integration to perform conduct prohibited here on your behalf. Rephrasing, encoding, translating, splitting, fictionalizing, or routing a request through multiple steps does not make prohibited conduct acceptable when the practical purpose remains the same.

We consider the entire context rather than isolated keywords. Legitimate analysis, journalism, education, historical study, safety research, defense, compliance, and creative work can discuss harmful subjects without meaningfully enabling harm. When a request can be safely answered at a higher level, the Service may provide prevention, recognition, recovery, or non-operational information instead of detailed instructions. Model refusals and safety transformations may occasionally be over- or under-inclusive.

2. Illegal conduct and serious harm

Do not use ZexirAI to plan, facilitate, conceal, coordinate, or profit from conduct you know is unlawful or that creates a clear and unreasonable risk of serious harm. This includes trafficking, kidnapping, extortion, stalking, burglary, arson, sabotage, organized violence, evasion of lawful sanctions, laundering criminal proceeds, obstruction of justice, destruction of evidence, and targeted assistance intended to help someone avoid detection while committing a crime.

You may ask for general legal information, harm-prevention guidance, safety planning, compliance assistance, or analysis of public cases. You may not use a generated explanation as permission to act. Laws vary across place and time, and ZexirAI does not verify your authority. When a request involves controlled access, surveillance, intrusion, weapons, personal records, or regulated activity, you must have a lawful purpose and every required authorization.

Do not use the Service to harass, threaten, intimidate, dox, shame, blackmail, exploit, or coordinate abuse against an individual or protected group. Content may criticize ideas or public conduct and may document abuse, but it must not meaningfully facilitate targeted harm. Credible threats and imminent danger should be reported to the appropriate local authority or emergency service.

3. Child safety

ZexirAI must never be used to sexualize, groom, exploit, traffic, threaten, or endanger a person under 18. Prohibited content includes child sexual abuse material; sexual or fetishized depictions of minors, whether real, generated, illustrated, or fictionalized; instructions for grooming or evading detection; sexual roleplay involving a minor; requests to estimate whether a young-looking person is legally old enough for sexual activity; or tools that facilitate access to children for exploitative purposes.

Do not upload, transform, summarize, classify, or reproduce suspected child sexual abuse material except through a legally authorized reporting or investigative system specifically designed for that purpose. Do not use ZexirAI as storage or as a substitute for the reporting channel. If you encounter such material, stop interacting with it and report it to the appropriate national hotline or law-enforcement authority.

Legitimate age-appropriate education, safeguarding, parental guidance, abuse prevention, and support for survivors are allowed when handled without explicit exploitative detail. Adults must not solicit a minor’s private contact information, precise location, intimate images, or secrets through content generated with the Service. Schools and youth programs must supervise use and apply required consent and privacy safeguards.

4. Sexual abuse and non-consensual intimate content

Do not create, alter, request, distribute, or threaten to distribute intimate or sexual content of a real person without their affirmative consent. This includes deepfakes, “nudification,” hidden-camera material, sexualized impersonation, revenge pornography, extortion, and attempts to identify anonymous people in intimate media. Consent to one image or context does not imply consent to alteration, publication, or another use.

Do not use the Service to facilitate sexual violence, coercion, exploitation, trafficking, or an inability to consent. Requests that eroticize intoxication, unconsciousness, force, dependency, or abuse are prohibited when they endorse or meaningfully enable non-consensual conduct. Clinical, legal, prevention, survivor-support, and news contexts may be discussed with care.

Adult sexual-health education and non-exploitative discussion may be allowed, but the Service is not a substitute for medical care or a guarantee that age, consent, or identity claims are true. Never use generated output to pressure someone, fabricate consent, misrepresent an identity, or evade platform safeguards.

5. Violence, weapons, and dangerous materials

Do not use ZexirAI for targeted instructions that would materially help a person injure or kill others, select vulnerable targets, optimize an attack, avoid detection, construct or acquire an illegal weapon, or produce dangerous biological, chemical, radiological, explosive, or toxic material. Do not seek operational assistance for terrorism, violent extremism, assassination, torture, mass casualty events, or attacks on critical infrastructure.

High-level education about history, public policy, physics, chemistry, medicine, risk recognition, emergency preparation, de-escalation, protective engineering, and lawful safety training can be appropriate. The line is crossed when detail, personalization, troubleshooting, or optimization makes harmful execution substantially easier. Professional work involving controlled hazards must occur in an authorized facility under competent supervision and applicable regulation.

Do not connect generated instructions directly to machinery, vehicles, laboratory systems, weapons, or safety-critical controls without validated engineering and authorized human review. If you find a suspicious device or hazardous substance, do not handle it based on AI instructions; leave the area and contact the appropriate emergency authority.

6. Self-harm, eating disorders, and crisis situations

Do not use the Service to encourage, romanticize, coach, optimize, or conceal suicide, self-injury, starvation, purging, or another life-threatening behavior. Requests for method comparison, lethal-dose calculation, evasion of intervention, or persuasive content that pushes a specific person toward self-harm are prohibited. The Service may redirect such requests toward immediate safety and human support.

Supportive conversation, recovery resources, coping strategies, safety planning, and educational information can be allowed. ZexirAI is not an emergency responder or therapist and cannot monitor a person’s condition. If you or someone else may act imminently, contact local emergency services or a crisis service and involve a trusted person who can be physically present. Verify local contact information from an official source.

Do not deploy ZexirAI as the sole crisis triage system, suicide-risk classifier, or replacement for trained care. Organizations using AI in a support context must provide qualified human escalation, appropriate notices, privacy protection, monitoring, and procedures for failures.

7. Cybersecurity and technical abuse

You may use ZexirAI for defensive security, authorized testing, secure development, education, capture-the-flag exercises, vulnerability remediation, malware analysis in a controlled environment, and other lawful work. You must own the system or have explicit permission covering the scope, timing, methods, and data involved. Follow coordinated disclosure rules and minimize access to personal or production data.

Do not use the Service to deploy or improve malware, ransomware, credential theft, phishing, destructive payloads, botnets, stealth persistence, unauthorized surveillance, data exfiltration, access-control bypass, exploit chains against real targets, denial-of-service attacks, supply-chain compromise, or techniques primarily intended to evade security detection while causing harm. Do not provide stolen credentials, live session tokens, private keys, or unlawfully obtained data for analysis or use.

Dual-use code may be limited when context indicates malicious intent, a real vulnerable target, broad scalability, concealment, or immediate operational capability. A defensive framing does not cure a request when the requested output is tailored for unauthorized exploitation. We may instead provide secure patterns, detection logic, sandboxed demonstrations, incident-response steps, or remediation guidance.

8. Fraud, impersonation, and deception

Do not use ZexirAI to steal money, credentials, identity, property, benefits, or services; forge records; create deceptive invoices; conduct phishing or social engineering; manipulate markets; evade know-your-customer controls; launder funds; submit false insurance, tax, immigration, academic, employment, medical, or government claims; or conceal who is behind a material transaction.

Do not impersonate a real person or organization in a way likely to deceive others about authorization, identity, origin, or endorsement. Satire, fiction, accessibility assistance, and authorized drafting may be allowed when the context does not create material deception. Synthetic audio, imagery, endorsements, testimonials, reviews, evidence, and public-interest communications must be labeled when law, platform rules, or reasonable audience expectations require disclosure.

Do not generate mass spam, fake engagement, coordinated inauthentic behavior, fraudulent reviews, astroturf campaigns, deceptive fundraising, or individualized manipulation at scale. Marketing and outreach must comply with consent, identification, opt-out, consumer-protection, and platform rules. Do not use sensitive traits or vulnerabilities to pressure people into consequential purchases, votes, donations, or disclosures.

9. Privacy, surveillance, and identity

Do not collect, infer, expose, purchase, combine, or use personal information without a lawful basis and appropriate notice. Do not dox people, reveal private contact or location information, identify anonymous individuals, track a person without authorization, build invasive profiles, or infer highly sensitive traits from appearance, voice, behavior, or location for consequential use. Public availability does not automatically make every use fair, safe, or lawful.

Do not submit another person’s health, financial, biometric, intimate, education, employment, communication, or government-identifier data unless you are authorized and the use is appropriate. Redact unnecessary information. Do not ask ZexirAI to discover passwords, recovery answers, private keys, authentication codes, or other access credentials. If credentials appear in Output or a file, stop using them and notify the owner through a safe channel.

Lawful security monitoring, fraud prevention, parental safety, accessibility, and public-interest research may involve personal data, but must be necessary, proportionate, accurate, and subject to safeguards. Facial recognition, emotion inference, biometric categorization, persistent surveillance, and location tracking require especially careful legal and ethical review and may be prohibited in some contexts.

10. Intellectual property and confidential information

Upload or request transformation of material only when you own it, have permission, or are otherwise legally entitled to use it. Do not use ZexirAI to distribute pirated works, bypass access controls or digital rights management, reproduce protected material beyond an applicable exception, remove ownership notices, counterfeit goods, or misappropriate trade secrets. A model’s willingness to produce content is not a license.

Do not prompt for a living artist’s exact protected work, a brand asset, or a person’s likeness in a way intended to deceive or substitute for a licensed original. Style, idea, and infringement rules vary, so commercial users should conduct appropriate clearance. Generated output may resemble existing material and may not be exclusive or protectable.

Honor professional and contractual confidentiality. Attorneys, clinicians, employees, contractors, students, researchers, and public officials may have duties that restrict use of third-party AI systems. Remove names and unique facts when possible, use an approved environment, and obtain required consent. Private mode does not create privilege, a nondisclosure agreement, or a regulated-data contract.

11. High-impact decisions and professional services

Do not use Output as the sole or determinative basis for decisions about another person’s employment, housing, education, credit, insurance, health care, legal services, immigration, criminal justice, access to essential services, or similar high-impact opportunity. AI can be inaccurate and discriminatory. Where any AI support is used, provide meaningful qualified human review, test for relevant errors and disparate effects, document the basis, give legally required notice, and provide an effective way to contest the result.

Do not falsely present ZexirAI as a licensed professional or use it to practice medicine, law, finance, engineering, or another regulated profession without appropriate human qualifications and oversight. Do not fabricate citations, diagnoses, credentials, signatures, inspections, audit findings, or certifications. A professional remains responsible for independently reviewing work and protecting client information.

Political persuasion, issue advocacy, and civic education must comply with election, campaign, advertising, privacy, and disclosure law. Do not create deceptive voting instructions, impersonate election officials, suppress participation, target political messages using unlawfully obtained sensitive data, or automate individualized manipulation that conceals its source. Verify election dates and procedures with official local authorities.

12. Regulated goods, gambling, and financial activity

Do not use ZexirAI to unlawfully buy, sell, manufacture, prescribe, distribute, or evade controls for drugs, weapons, explosives, restricted chemicals, counterfeit documents, stolen goods, endangered species, or other regulated products. General education, compliance, harm reduction, and lawful professional workflows may be appropriate when they do not materially facilitate evasion or dangerous use.

Do not use the Service to operate illegal gambling, manipulate a market, trade on misappropriated inside information, evade sanctions or anti-money-laundering controls, or provide personalized regulated financial services without authorization. Output about markets is informational and may be wrong or outdated. Verify information and use a qualified professional before taking financial action.

Age-restricted products and activities require reliable age and location controls outside the model. A user’s statement about age or jurisdiction is not sufficient assurance for a regulated transaction. ZexirAI must not be used as the only compliance or identity-verification system.

13. ZexirAI service integrity

Do not interfere with the Service or other users. Prohibited conduct includes bypassing authentication, safety controls, rate limits, subscription gates, quotas, storage limits, or geographic restrictions; using unauthorized automation or scraping; creating accounts to evade enforcement; probing or exploiting vulnerabilities without permission; overwhelming infrastructure; injecting malicious files; manipulating usage records; or accessing records that do not belong to you.

Do not reverse engineer, extract, or replicate protected models, prompts, datasets, interfaces, or systems except to the limited extent applicable law expressly permits. Do not use Output or automated access to build a competing model or service where the Terms, technical controls, or applicable provider conditions prohibit it. Ordinary interoperability, accessibility, and authorized security research remain subject to written scope and law.

Plan capacity is for the account and authorized use described at purchase. Do not resell, pool, lease, or broker access, or operate an undisclosed public proxy. Do not share individual credentials to avoid purchasing appropriate access. We may throttle unusual activity, require reauthentication, or request information needed to distinguish legitimate high-volume work from abuse.

14. Enforcement, appeals, and reporting

We may use automated signals and human review to detect or investigate potential violations. Depending on severity, context, history, and urgency, we may warn, limit output, block a request, remove content, throttle access, revoke sessions, restrict a feature, suspend or terminate an account, preserve evidence, notify a provider, or report conduct when law requires. We may act immediately when delay could expose a person or the Service to serious harm.

Enforcement systems can make mistakes. If you believe a restriction was incorrect, contact support with the account email, approximate time, affected feature, and enough non-sensitive context to review the decision. Do not resend prohibited content or attempt to evade the restriction. We may uphold, modify, or reverse the action and may decline repetitive or abusive appeals.

Report harmful use or a policy concern to mail@zexiry.com. For immediate danger, contact the appropriate emergency service. We may update this policy as capabilities, misuse patterns, provider requirements, and laws evolve. The effective date identifies the current version, and material changes receive notice when required.

1. Lawful Use: detailed guidance

This section explains how the policy applies to lawful use. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on lawful use, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to lawful use. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns lawful use, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of lawful use as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

2. Personal Safety: detailed guidance

This section explains how the policy applies to personal safety. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on personal safety, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to personal safety. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns personal safety, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of personal safety as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

3. Violent Wrongdoing: detailed guidance

This section explains how the policy applies to violent wrongdoing. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on violent wrongdoing, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to violent wrongdoing. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns violent wrongdoing, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of violent wrongdoing as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

4. Self-Harm Response: detailed guidance

This section explains how the policy applies to self-harm response. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on self-harm response, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to self-harm response. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns self-harm response, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of self-harm response as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

5. Sexual Exploitation: detailed guidance

This section explains how the policy applies to sexual exploitation. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on sexual exploitation, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to sexual exploitation. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns sexual exploitation, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of sexual exploitation as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

6. Minor Safety: detailed guidance

This section explains how the policy applies to minor safety. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on minor safety, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to minor safety. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns minor safety, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of minor safety as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

7. Harassment: detailed guidance

This section explains how the policy applies to harassment. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on harassment, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to harassment. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns harassment, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of harassment as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

8. Hate And Discrimination: detailed guidance

This section explains how the policy applies to hate and discrimination. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on hate and discrimination, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to hate and discrimination. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns hate and discrimination, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of hate and discrimination as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

9. Privacy Violations: detailed guidance

This section explains how the policy applies to privacy violations. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on privacy violations, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to privacy violations. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns privacy violations, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of privacy violations as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

10. Identity Abuse: detailed guidance

This section explains how the policy applies to identity abuse. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on identity abuse, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to identity abuse. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns identity abuse, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of identity abuse as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

11. Deceptive Media: detailed guidance

This section explains how the policy applies to deceptive media. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on deceptive media, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to deceptive media. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns deceptive media, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of deceptive media as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

12. Fraud And Scams: detailed guidance

This section explains how the policy applies to fraud and scams. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on fraud and scams, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to fraud and scams. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns fraud and scams, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of fraud and scams as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.

13. Malware And Intrusion: detailed guidance

This section explains how the policy applies to malware and intrusion. The governing principle is that a user should understand the feature, use only the access and information reasonably needed for the task, and keep qualified human judgment in control whenever an action may affect another person. Product labels, examples, interface text, and automated notices illustrate ordinary operation; they do not replace the specific facts, laws, contracts, professional duties, or provider terms that apply in a particular situation.

Before relying on malware and intrusion, review the requested purpose, the sensitivity of the information involved, the people who may be affected, and the consequences of an error. Use the least sensitive input that can complete the task. Confirm names, dates, amounts, permissions, sources, and destinations before sending, publishing, purchasing, deleting, or connecting anything. If a result appears incomplete, inconsistent, unusually confident, or outside the expected context, pause and verify it through an independent and authoritative source.

ZexirAI may apply technical limits, safety checks, rate controls, account verification, logging, provider routing, and human review to support this policy in relation to malware and intrusion. These measures reduce risk but cannot identify every mistake or misuse. A control that does not block an action is not approval, legal advice, a guarantee of accuracy, or a waiver. Users remain responsible for instructions they provide, content they choose to keep, actions they authorize, and the final use of any output.

When an exception, dispute, or unexpected result concerns malware and intrusion, preserve the minimum information needed to understand what happened: the approximate time, affected account, feature used, relevant settings, safe transaction or request identifiers, and a concise description of the expected and actual behavior. Remove passwords, one-time codes, full payment details, unrelated personal records, and confidential material. Contact support promptly when delay could increase harm, and use emergency or official reporting channels when the situation requires them.

We review the operation of malware and intrusion as the Service, models, providers, security threats, and legal requirements change. We may clarify instructions, adjust safeguards, restrict a capability, or request additional verification. Material policy changes receive notice when required. Existing non-waivable rights continue to apply. If this detailed guidance conflicts with a more specific section of this policy, the more specific section controls for that issue; the Terms and applicable law control the overall relationship.