Privacy Policy

1. Who we are and when this Policy applies

SourceWarden LLC, a Pennsylvania limited liability company, operates EvidenceGuard. Our contact is kun11451454188@gmail.com; postal contact is SourceWarden LLC, 7155 Kindred Street, Philadelphia, PA 19149-1132, United States. Identify privacy requests as “EvidenceGuard privacy” and do not include passwords, tax IDs, authentication tokens or confidential documents in an initial message.

This Policy describes personal information processed in connection with our website, accounts, communications and Service. This text is version 1.0, issued 4 October 2026. It applies when SourceWarden publishes it for the relevant Service; the live notice’s publication record identifies when it takes effect. It does not retroactively authorize a new use of information. Contractual acceptance and changes to the Terms are governed separately by the Terms of Service.

We determine purposes for account administration, billing, support, necessary operational security and our own legal obligations. When a customer supplies documents for processing on its behalf, that customer generally determines the purpose and acts as controller or business, and we act as processor or service provider. If the customer itself processes for another organization, we act in the corresponding subprocessor role. The DPA governs personal data in Customer Content; we do not relabel that content as our own data to reuse it independently.

An individual mentioned in a customer document should normally contact that customer about its use of the document. We will route or assist with a request and meet any duty applicable to us rather than leave it unanswered. Customer privacy notices and third-party websites may separately apply to their activities.

2. Information we collect or receive

Account and organization information. Names, work email addresses, sign-in identifiers, organization details, membership and permissions, invitation and acceptance records, preferences and authentication-related records. Authentication providers process credentials or session information needed for their role; our application does not need you to send a password to support.

Documents and review information. Authorized PDFs, associated metadata, text or page representations produced during processing, source and version references, observations, requirements, findings, review notes and human decisions. Documents can contain ordinary business contact details and professional information about employees, suppliers, consultants or public officials. Derived content remains subject to the same purpose restrictions as its source.

Activity and audit records. Organization and actor identifiers, timestamps, permissions, policy changes, document or artifact references, recorded authorizations and review actions. These support access control and accountability. They are not necessarily anonymous and are not optional advertising statistics.

Contact and support information. Messages, contact details and minimal technical evidence voluntarily provided in inquiries, support requests or complaints. Do not include unnecessary sensitive information. We arrange a suitably limited channel and permission before requesting access to confidential content.

Transaction information. Where a paid service is actually offered, Orders, invoices, grants, credits, consumption, refund records, payment status, transaction identifiers and limited payment-method information supplied by the processor. Payment credentials are submitted to the authorized processor through its payment flow, not to ordinary support. A sandbox integration does not mean live payments are available.

Technical information. Information needed to deliver requests and maintain the Service, such as network address, request time, route or request path, response status, browser or device information, referrer where transmitted, authentication events and service diagnostics. Depending on the infrastructure layer, an address may identify a proxy rather than a person. Infrastructure logs are separate from minimized browser diagnostic reports.

Optional product-statistics contributions. When the applicable settings and legal conditions permit storage, the system can store organization context, day, feature category, a daily user-derived identifier and an expiry time. These contributions are pseudonymous, not described as anonymous. The contribution record does not contain source text or a source locator. An asset reference may be transmitted to check authorization and eligibility without being stored as a contribution field.

We do not request sensitive identity records, regulated health or student data, payment authentication secrets or other restricted categories through the standard document service. Incidental receipt does not authorize unrestricted retention or use. Contact us about an accidental submission so it can be handled appropriately.

3. Sources and purposes

We receive information from you and your organization, from documents your organization authorizes, from sign-in and payment providers used for the requested function, and from technical activity involving the Service. Public product demonstrations use illustrative content, not a visitor’s confidential document. We do not purchase third-party advertising profiles for the purposes described here.

We use appropriate information to establish accounts and permissions; deliver approved document processing and source access; maintain review and version context; operate and protect the Service; diagnose failures; respond to inquiries and rights requests; administer accepted payments and promotions; prevent abuse; enforce our agreements proportionately; and comply with applicable legal obligations.

Optional product statistics are used to understand use of identified features and improve their operation, not to rank individual employees or infer confidential bid content. A human correction, accepted finding or support conversation is not automatically a training example. We do not treat a broad “improve the service” purpose as permission to collect a customer-content research corpus.

Where we process a customer’s documents as a processor, those operations follow the customer’s documented instructions and the DPA. Personal information necessary for our own billing, fraud prevention or legal compliance is separately limited to those purposes.

4. Legal bases where required

Where a law requires a legal basis for our controller activities, we rely on the basis applicable to the particular purpose. Processing may be necessary to perform a contract with an individual who is our customer, take requested pre-contract steps, comply with a legal duty, or pursue a legitimate interest in providing business communications, secure accounts, fraud prevention and a functioning Service where that interest is not overridden by individual rights. An agreement with a company is not automatically a contract with every employee or person named in a document.

We rely on consent where required for an optional purpose or communication. Consent can be withdrawn for future processing without affecting processing lawfully performed beforehand. Where legitimate interests apply, a right to object may be available. Browser storage and electronic communications may also be subject to separate rules. We assess those conditions before enabling a new nonessential activity; an organization setting is not a substitute for an individual’s required choice.

5. No sale, behavioral advertising or default training

We do not sell personal information or share it for cross-context behavioral advertising, and we do not use Customer Content to build advertising profiles. We do not provide Customer Content to another party for its independent marketing or general-purpose model training.

We do not train or fine-tune a general-purpose model on Customer Content or create an independent research dataset from it under this Policy. The baseline does not offer a customer-content research program. A separately proposed purpose would need a specific notice, suitable authority and any required consent or agreement before processing; merely changing this Policy or accepting the Terms would not authorize it.

Customer-document external AI is disabled in the baseline covered by this version. A future enabled external-processing feature must identify its actual provider, content scope, purpose, data handling and authorization conditions under the AI Addendum. We do not claim a provider-specific no-training, zero-retention or residency arrangement without the actual arrangement being established.

6. Cookies, browser storage and diagnostics

The Service uses browser storage for requested functions, including authentication state, organization selection and an allowed sign-in return destination. Clearing it can sign you out or reset the workspace selection; it does not delete server records. Local storage does not necessarily have a fixed automatic expiry.

The first-party browser fault reporter uses limited categories and coarse application areas, with a status or constrained request reference where available. Its event payload is not a screen recording and does not capture document text, DOM contents, exception messages or exception stacks. The request still involves authentication and organization context, and separate server and provider logs may process additional metadata. This is not a promise that all logging is anonymous or that no request URLs exist in infrastructure logs.

The organization product-statistics policy defaults to off. Some feature paths may still send an eligibility request that the server suppresses rather than stores; a successful HTTP response does not prove that a contribution was recorded. Where stored, contributions have a 30-day lifecycle and reports suppress small groups and round counts. Turning the setting off affects new eligible contributions, not every diagnostic record or immediate deletion of earlier records.

We do not deploy cross-site advertising or session-replay collection as part of this baseline offering. We do not use a browser “Do Not Track” signal to disable necessary authentication, security or service records. Our no-sale and no-cross-context-sharing policy applies whether or not a Global Privacy Control signal is present. We will honor an applicable opt-out signal for a legally covered activity rather than treat it as consent; this does not imply a control that disables all necessary processing. More detail appears in the Telemetry and Browser Storage Notice.

7. Recipients and disclosures

Authorized personnel and providers receive only information needed for their permitted role. The Service Provider and Subprocessor Directory identifies infrastructure and other providers and distinguishes customer-content hosting from authentication, DNS, communications and conditional payment services. The documented architecture uses DigitalOcean for application hosting and managed business data, AWS S3 for private object storage, and Supabase for authentication. An installed self-hosted parser or scanner is not itself an external recipient merely because another company develops it.

We disclose information to Customer’s authorized administrators and users as directed by its workspace permissions; to relevant service providers; to professional advisers under appropriate duties; and where legally required or reasonably necessary to address unlawful activity or protect rights and security. We assess such requests, limit disclosure to what is necessary and notify affected customers where lawful and appropriate.

A corporate transaction may involve limited due diligence or transfer of business records under appropriate confidentiality and data-protection safeguards. The recipient remains subject to applicable commitments; a transaction does not create a new unrestricted use of Customer Content. We do not list a bank or a developer tool as a document subprocessor merely because SourceWarden uses it for its own business.

Customer-selected sign-in, payment, export or integration services may have independent responsibilities and notices for their own activities. Customer must authorize any export it requests. The fact that a provider publishes a DPA is not proof that every possible service, affiliate or region is authorized for your documents.

8. Processing locations and international safeguards

The standard application and document-storage deployment is based in the United States. Provider account administration, authentication support, communications, network operations and authorized support may involve other countries. A United States storage region is not a promise that every service record or support operation stays in the United States.

Before accepting processing that requires an international-transfer safeguard or a particular local environment, we will identify and establish the necessary arrangement. Depending on the law and actual recipient, this may involve an applicable adequacy determination, valid standard contractual clauses or another legally available mechanism, together with required assessments and supplementary protections. We do not represent that SourceWarden is certified under a Data Privacy Framework or that its own DPA is an official transfer instrument. Request information about the arrangement applicable to your data through the privacy contact.

The standard beta must not be used to initiate a restricted international transfer of customer documents before the required arrangement is in place. This limitation does not remove privacy rights that already apply to website or account processing.

9. Retention schedule

We retain information only for its stated purpose and justified legal or security needs. The periods below are the retention rules for processing governed by this version; an applicable law or a more protective accepted arrangement prevails. A documented preservation duty can extend a specific record’s period, but not all records indiscriminately.

Raw uploads and content-bearing derivatives. The organization’s actual 7-day, 30-day, 90-day or indefinite raw-upload setting governs scheduled source retention; new organizations default to 30 days and some older organizations retain an indefinite setting until changed. Content-equivalent extracted text and page material follow the source-related instruction unless a separately authorized retention exception applies. Indefinite does not authorize another processing purpose. A verified deletion or closure request triggers the separate cleanup process described below.

Optional product contributions. Contributions expire after 30 days and are excluded from eligible reporting when expired. Cleanup removes expired primary records. This expiry is not a representation of instantaneous removal from every backup.

Routine operational and diagnostic logs under our control. We use bounded rotation for ordinary application and container logs and retain them only for operational, security and troubleshooting needs. Retention can vary with log volume, infrastructure layer and provider-controlled schedules, so this Policy does not assign one fixed day count to every log source. We minimize content in routine diagnostics and review retained logs rather than treating them as an indefinite archive. Records isolated for a genuine security incident may be kept longer where reasonably necessary for investigation, remediation, legal obligations and accountability, with only necessary content retained.

Account profiles and organization membership. Active account information is held while needed to administer the account. Deactivation revokes application access before every underlying record is necessarily removed. Following a verified closure or deletion request, unnecessary profile and membership information is removed or de-identified through the account-closure process, while separately justified transaction, consent, security and audit records may remain for their stated purposes. We do not treat deactivation alone as proof of complete physical deletion.

Customer audit metadata. We retain minimal action and authority records while needed to operate the workspace, investigate security or access issues, demonstrate material authorizations, resolve disputes, and meet applicable legal or accountability obligations. We periodically review retained audit data and minimize unnecessary personal identifiers. Content-bearing passages are not entitled to longer retention merely because an audit record or historical report exists.

Agreements, billing and tax records. We retain the records reasonably needed to substantiate contracts, payments, credits, refunds, disputes and relevant tax or accounting obligations for the period required by applicable law and reasonably necessary business recordkeeping. We apply the relevant record type and legal purpose rather than using this category to extend retention of unrelated uploaded bid materials.

Ordinary support and inquiries. We keep routine support and business-inquiry correspondence only while reasonably needed to resolve the matter, maintain appropriate business records, or meet a legal, privacy or security obligation. Separately authorized confidential diagnostic samples are removed when the support task no longer requires them, subject to a documented preservation duty. Ordinary email is not an approved archive for Customer Content.

Privacy requests and communication preferences. We retain the minimum record needed to demonstrate handling of a privacy request, honor a continuing preference, resolve a dispute, or meet an applicable legal obligation. We periodically review those records and remove unnecessary verification material. A minimal suppression record may be kept as long as needed to honor a requested communication opt-out; it is not permission to keep a marketing profile.

Restricted backups and recovery copies. Recovery copies are separate from ordinary application access and can retain historical versions after primary-system deletion. They are used for recovery, not analytics or independent reuse. A complete-deletion request therefore distinguishes active-system deletion from backup handling. We restrict recovery-copy access, track applicable deletion state, and require a restore process to reapply deletion and access restrictions before ordinary use resumes. We do not claim that a primary delete instantly removes every historical backup copy or that a backup copy has been erased until that cleanup or expiration has actually been verified.

10. Deletion is not one action

Deleting a source revokes normal source access and queues object purge. Storage errors or pending jobs are not completed deletion. Raw-source deletion, deletion of content-bearing derivatives, removal of optional contributions, user deactivation and organization closure can require distinct steps. Organization closure is support-assisted in the beta; there is no promise of a universal one-click erase of every database and backup.

For a verified complete-content deletion instruction, we remove applicable data from active systems without undue delay and track asynchronous purge work until it is complete. We explain a technical impediment or lawful exception and complete the remaining work promptly, subject to mandatory deadlines. Necessary limited records and restricted backups follow the separate rules above. We do not represent a queued purge, revoked access, or primary-system deletion as completed physical erasure of every recovery copy.

A lawful hold must identify its basis, scope and review date. We restrict the retained data, do not use it for unrelated purposes, and resume deletion when the justification ends. Customers should retrieve needed originals and outputs before their chosen retention expires. A request after expiry cannot recreate data already properly deleted.

11. Individual rights and requests

Depending on applicable law and our role, you may request access, correction, deletion, portability, restriction, information about recipients, withdrawal of consent, or objection to certain processing. You may have rights concerning sale, sharing, targeted advertising, sensitive-data use or consequential profiling. We do not perform those advertising or decision-making activities under this baseline. We will not retaliate or unlawfully discriminate for exercising a right.

Send your request to the privacy contact above. State the relationship to the Service, the request and enough information to locate relevant records without sending an unnecessary identity document. We use proportionate verification. An authorized agent may make a request with the authority required by applicable law; we will explain any verification needed. We do not disclose another person’s or organization’s confidential information merely because a request is made.

We aim to acknowledge requests within 10 business days and respond within 30 calendar days, or earlier if law requires. Any lawful extension, refusal or limitation will be explained with the applicable reason. Where Customer controls the data, we will coordinate with it and assist rather than make an unauthorized substantive decision. Necessary action is not conditioned on buying a subscription or waiving a right.

To appeal a privacy-request decision where applicable, reply to the decision or email the same address with subject “Privacy appeal.” We will reconsider and provide a reasoned response within 45 days or any shorter required period, explain any lawful extension, and identify an available regulator complaint route where required. You may complain directly to the relevant supervisory authority or state attorney general without first appealing to us.

12. California and other United States residents

To the extent applicable state privacy law covers the processing and individual, the categories in Section 2, purposes in Section 3, recipients in Section 7 and retention rules in Section 9 describe our practices. These include identifiers and contact details, commercial transaction information, internet or network activity, professional and employment-related information, and customer-provided content. We do not use sensitive personal information for unrelated inference, targeted advertising or a separately disclosed restricted purpose under this offering.

We disclose necessary information for the business purposes described, not for sale or cross-context behavioral advertising. We do not knowingly sell or share personal information of persons under 16. The request procedure in Section 11 accommodates applicable access, correction, deletion, portability, agent and appeal rights. A statutory right applies subject to the law’s definitions, scope and exemptions; this Policy does not claim that SourceWarden meets every state’s business threshold or that business-contact information is exempt everywhere.

We update the notice when practices materially change and review it at least annually. Prior versions are available on request. We do not report a fictional 12-month operating history for a newly established service.

13. Children and automated decisions

EvidenceGuard accounts are for adult business users. We do not knowingly solicit children’s information for the standard service or design the public example for children. An adult-only account policy does not prove that every customer document lacks information about minors. Customers must keep restricted categories out of the standard service. Contact us about an inappropriate submission so we can investigate and take appropriate action.

Automated processing supports a human document-review workflow. We do not offer the Service to make solely automated decisions with legal or similarly significant effects about individuals. A user clicking an approval button is not by itself proof of meaningful human review. Uses requiring a different assessment are outside the standard scope.

14. Security, changes and contact

We use organization-scoped access, private storage, secure transport, controlled document processing, credential protection and relevant diagnostics as safeguards, with the minimum commitments in the DPA. No system is completely secure. We do not claim an independent SOC 2 or ISO certification or that a source hash or malware scan proves a document safe or correct.

We will communicate material changes through the website and, where appropriate, account email or in-service notice. We will not use a changed notice as retroactive consent to a new purpose. Separate approval will be obtained where required. The public notice will identify its actual publication/effective date and version; the issuance date of prepared text alone is not a record of publication.

Privacy, data-return or security inquiries: kun11451454188@gmail.com. Mail: SourceWarden LLC, 7155 Kindred Street, Philadelphia, PA 19149-1132, United States. Please request a suitable secure channel before sending confidential material.