Content Moderation & Takedown Policy

Effective date: 6 November 2026

1. Scope and principles Lighthouse reviews profiles, portfolios, products, reviews, advertisements, RFQs, uploaded files and enabled AI features for violations of law and platform rules. We seek proportionate decisions, clear reasons and meaningful review. Paid status does not buy exemption from enforcement. A moderation decision is not a judicial determination of ownership, contractual liability or professional negligence. We do not promise that every item is checked before publication or that automated tools detect every violation.

2. Grounds for action Action may address impersonation, false credentials, stolen or fabricated projects, counterfeit or unsafe goods, misleading prices or stock, manipulated reviews, spam, harassment, privacy invasion, intellectual-property infringement, unlawful content, malicious files, fraudulent RFQs, collusive quotations and deceptive advertising. Public project showcases are limited to real completed projects. AI or conceptual material must not be used as false evidence of completed work. We distinguish good-faith routine editing from deceptive synthetic content and preserve required labels. A genuine negative review, criticism or commercial disagreement is not automatically prohibited. We assess context, evidence, public interest and lawful exceptions.

3. How to report Use [CONFIRM reporting URL] or the grievance contact in the Contact, Grievance & Dispute Resolution Policy. Identify the exact profile, project, listing, review or file; explain the concern; provide relevant evidence and a contact for follow-up. Anyone may report a safety or rule concern; rights claims should be made by the rights holder or an authorised representative. Do not upload unnecessary identity documents or repeatedly transmit intimate or exploitative material. Give a locator and a minimal description through the secure route. Urgent physical danger should also be reported to emergency services. Copyright and trademark reports follow their respective policies. We seek missing information without allowing clarification requests to defeat mandatory timelines.

4. Intake, assessment and interim measures We record receipt, allocate a reference, classify urgency and route the case to an authorised reviewer. We assess the precise content, rule, evidence, potential harm and applicable legal process. We may ask either party for a response or evidence such as licences, original files or order records.

Pending a decision, proportionate measures may include a warning label, restricted access, a listing pause or temporary removal. We do not disclose private RFQs or full legal evidence to unrelated users. Records needed for investigation are preserved securely.

5. Applicable complaint and removal clocks The Contact, Grievance & Dispute Resolution Policy contains the central response-clock table. For intermediary matters under the current Indian rules, the ordinary route acknowledges within 24 hours and resolves within seven days; specified prohibited-content removal complaints have a 36-hour resolution limit. Complaints falling within the intimate-image/sexual-content or electronic-impersonation category require reasonable and practicable removal/disablement measures within two hours. Qualifying court orders or authorised government intimations under rule 3(1)(d) require removal/disablement within three hours. These categories have distinct legal triggers; they are not a promise to remove every allegation within two or three hours. Statutory clocks run from the applicable receipt/knowledge event, including outside business hours. Stricter valid directions control. Consumer order complaints follow the separate consumer route unless another shorter requirement also applies.

6. Decision and notice Available outcomes include no action, correction, attribution repair, labelling, feature restriction, demotion, removal, temporary suspension or termination. We consider severity, reliability of evidence, repeat behaviour, remediation and proportionality. Where lawful, we tell the affected user what was acted on, the rule or legal basis, the scope/duration and how to appeal. We notify the complainant of the outcome to the extent appropriate. We may withhold details to protect a victim, confidential information, an investigation or a legal restriction.

7. Appeals and counter-statements An affected user or complainant may seek human review through the grievance channel, quoting the case reference and explaining the error or supplying new evidence. A different reviewer should assess an appeal where practicable. Proposed internal submission window: 30 days after notice; this does not cut down a statutory appeal period or justify refusing a justified late request. A counter-statement can show authorisation, lawful use, mistaken identity or factual correction. It does not automatically restore content. We consider lawful exceptions and any controlling order before restoration. An internal appeal does not suspend an urgent statutory takedown or replace a court remedy. Where eligible, a person may use the Grievance Appellate Committee route under applicable law; the central grievance policy explains that external option.

8. Repeated and abusive conduct Repeated substantiated violations may lead to escalating restrictions, including linked accounts used to evade enforcement. We consider seriousness rather than operate an unexplained automatic strike count. False complaints, fabricated evidence and retaliatory reporting may themselves lead to action. Disagreement with a decision is not by itself abusive. A paid rights-holder programme, if introduced, must not disadvantage urgent statutory complaints by non-paying users.

9. Automated review and synthetic information Automated tools may assist identification or prioritisation but can make mistakes. Consequential disputed decisions must have a human-review route. We document the use and limitations of tools and avoid claiming universal proactive monitoring. The AI Policy governs safeguards for unlawful synthetic information, visible disclosure and technically feasible provenance. Additional significant-social-media-intermediary duties apply only if Lighthouse meets that legal classification; the classification must be periodically assessed.

10. Evidence, disclosure and transparency Access to complaint records is limited to authorised personnel and lawful recipients. Mandatory preservation and legal holds follow the Data Retention Schedule, including applicable 180-day intermediary preservation obligations. Material retained for evidence is not available for marketing or general model training. We assess the authority and scope of government requests and preserve records of the response. We provide legally required user notices and transparency information. No policy grants unlimited power to expose a complainant’s identity to another user.

Implementation checks Appoint the grievance officer, establish monitored urgent intake, build content-report and user-blocking controls appropriate to the app, maintain an escalation rota, and log receipt/decision clocks. Confirm legally required local-language access and periodic user reminders before publication.

11. Decision categories and proportional measures We distinguish unlawful content, content that violates a published platform rule, unverified claims requiring correction and material that is merely unpopular or critical. Moderation may include a warning, correction request, label, visibility limit, feature restriction, removal or account suspension, according to severity, recurrence, evidence and potential harm. Paid status, commercial influence or pressure from a featured firm must not determine whether a valid complaint is investigated. We may act before notifying an uploader where necessary to prevent harm or comply with law, but should explain the action later when lawful.

12. Reports involving historic projects and imports Identify whether the disputed asset was uploaded by the contributor, imported by Lighthouse staff, generated by an AI tool or embedded from another service. Lighthouse must investigate its own editorial/import decisions rather than treating every issue as solely the uploader’s responsibility. Check the source permission, final publication approval, client privacy, photographer rights and current accuracy of professional credits. Where permission is uncertain, restrict the affected asset pending review. Do not require the rights holder to create a commercial account or first contact the featured architect. Restrictions must reach controlled recommendations, search, thumbnails and scheduled posts. A removed photograph must not remain visible in an AI-generated carousel or legacy campaign simply because the main project page was disabled.

13. Automated detection and human review Automated systems may flag duplicates, unsafe content, false credentials or abuse, but can make mistakes. Material restrictions should have the human review required by law and this policy. Reviewers must consider context, language, satire, evidence and legitimate exceptions rather than relying solely on a keyword or score. A reason notice should identify the restricted material, the rule or legal ground, territorial scope, duration, whether automation materially informed the decision and the available challenge route, to the extent required and lawful. Do not disclose information that would compromise safety or lawful investigations.

14. EU Digital Services Act supplement For services within scope, provide an accessible electronic mechanism for notices with a sufficiently substantiated explanation, precise location, required contact information and good-faith accuracy statement, subject to statutory exceptions. Acknowledge receipt and communicate a decision as required. Prioritise trusted-flagger notices where the relevant provisions apply. Provide statements of reasons for covered restrictions and a complaint mechanism meeting the applicable DSA requirements. Where Article 20 applies, allow the required period of at least six months, free of charge, and qualified human supervision of decisions. Explain relevant certified out-of-court dispute settlement and judicial routes. Assess the applicable duties and any micro/small-enterprise exemptions by the actual service. Keep required transparency reporting and regulatory contact arrangements. Do not claim that every DSA obligation applies identically to a small host and a designated very large platform.

15. UK and other international services Before enabling a UK user-to-user service, determine whether the Online Safety Act and Ofcom requirements apply, including risk assessment, complaints and child-access

assessment. An 18-plus term alone does not establish that children cannot access the service. Apply local requirements without incorrectly describing the EU DSA as UK law. US copyright complaints follow the distinct DMCA procedure where activated. Other unlawful-content, privacy and safety reports are assessed under applicable federal/state rules and platform policies. Geo-restriction may be appropriate for a territory-specific order, but a global removal requires its own basis.

16. Abuse of reporting, evidence and transparency Repeated manifestly unfounded reports, forged court orders and coordinated harassment through reporting tools may lead to proportionate restrictions following assessment. Do not punish good-faith complaints solely because they are not ultimately upheld. Maintain restricted evidence of the report, relevant content version, decision, action and review. Record corrections and reversals. Retention must follow the applicable statutory period and legal holds, not a blanket indefinite archive. Summarise enforcement information where legally required without exposing victims or sensitive security methods.

17. Moderation case record and classification Each substantive case should identify the reported material, report source, receipt time, alleged issue, applicable policy or legal basis, urgency and reviewer. A report may concern several issues, such as copyright, privacy and counterfeit goods; each material basis must be considered rather than closed under an unrelated category. Lighthouse will distinguish removal of content, restriction of visibility, disabling a transaction, withdrawal of a badge and suspension of an account. These actions have different effects and should not be described interchangeably. The applicable legal clocks remain those stated in this policy and the grievance schedule.

18. Evidence and interim measures Relevant evidence may include the reported record, original source, authorisation, transaction history and the parties' explanations. Automated risk scores and report volume may help prioritise review but do not themselves prove a violation. Coordinated reporting must not automatically outweigh credible contrary evidence. A temporary restriction may be used where the risk justifies action before a full response. Its scope should be limited where feasible and periodically reviewed. A restriction of one disputed image need not disable every unrelated lawful project, unless the account-level risk warrants it.

19. Reasoned notices and lawful limits Where notice is required or appropriate, explain the material acted on, nature and territorial scope of the restriction, basis, relevant use of automation and available challenge route. Do not provide a generic “community violation” explanation where a more specific lawful explanation is practicable. Notice may be delayed or limited where law prohibits it or disclosure would create a credible safety or investigation risk. Record the reason internally and revisit the restriction when

appropriate. Confidentiality must not be used merely to avoid explaining an inconvenient commercial decision.

20. Human review and correction A reviewer must have access to relevant evidence and authority to change the outcome. Review should examine whether the correct rule was applied, the evidence supports the finding, the measure is proportionate and new information changes the decision. Payment status and business relationships must not influence the standard. If a decision is reversed, restore the affected lawful functionality where practicable and correct associated strikes, badges, rankings or notices that depended on the error. Restoration must still respect a separate continuing legal restriction. Explain any part that cannot be restored and why.

21. Territorial restrictions and legal orders A binding order must be checked for authenticity, scope, authority, affected material and any territorial limit, while meeting applicable deadlines. Lighthouse may seek clarification or challenge an overbroad demand through lawful means without delaying compliance where required. A local restriction should not automatically become global unless the order, applicable law or independently justified platform policy requires that result. Geographical limits must not be presented as a guarantee that every copy elsewhere on the internet is removed.

22. Reuploads, archives and derivative material Where the basis for removal applies to controlled copies or derivatives, the restriction should propagate to thumbnails, feeds, search, recommendations and AI summaries. Reupload prevention should account for lawful revised versions and mistaken matches. A shared visual hash alone may be insufficient to distinguish prohibited and lawful contexts. Evidence preserved for a dispute must remain separated from ordinary content delivery. Backup restoration, scheduled content import and catalogue refresh must not automatically republish restricted material.

23. Reporter abuse and transparency Knowingly false claims, threats, purchased reports and repeated misuse of the reporting process may be restricted proportionately. An unsuccessful good-faith complaint is not automatically abusive. Complainants retain available external remedies. Lighthouse should maintain information needed for applicable transparency duties and internal quality review, while avoiding disclosure of private complainant details or security-sensitive detection methods. Public statistics, where issued, must distinguish reports, decisions, appeals and reversals rather than equate every report with a proven violation.

24. In-app reporting and blocking requirements Public project stories, profiles, reviews and other user submissions require ongoing moderation even when chat is disabled. Before creating or uploading content, users must accept the applicable user rules. Provide in-app reporting for content and accounts and user-blocking controls appropriate to public content and direct interaction. A support email is a fallback, not a replacement for required in-app controls. Reporting must identify the relevant item without exposing the reporter publicly. Blocking and filing a report are separate actions.

25. Blocking scope and transaction continuity Explain what blocking changes in the implemented product: contact requests, mentions, notifications, profile interaction and content visibility. Do not promise prevention of all off-platform contact or every logged-out view. Blocking must not destroy evidence or remove access to order remedies. For an active dispute involving a blocked supplier, provide a support-mediated route rather than require renewed personal contact. Attempts to evade a block through another account may lead to restrictions after assessment.