Privacy Notice

PublicVoiceShield is operated by Mohammad Asif, trading as PublicVoiceShield, based in the United Kingdom. Until a registered company is formed, Mohammad Asif acts as the data controller for account, billing and service-operation data.

A future company entity may take over this role. Placeholders are shown below and will be completed at that point: future company name [Future company name — TO BE CONFIRMED]; future company number [Future company number — TO BE CONFIRMED]; future ICO registration number [Future ICO registration number — TO BE CONFIRMED].

Where your organisation submits content about other people (for example public comments), your organisation will usually be the controller of that content and PublicVoiceShield acts as a processor on your instructions. Roles depend on how you use the service — see the Draft Data Processing Agreement.

• Privacy contact email: privacy@publicvoiceshield.com
• Business contact address: [Business contact address — TO BE CONFIRMED]
• No Data Protection Officer has been appointed at this stage.

• Account details, including your name and work email address.
• Organisation membership, role and invitation status.
• Authentication and security information, such as sign-in events and session records.
• Original posts and pasted comment threads submitted for analysis.
• AI analyses, scores, categories and suggested replies.
• Human decisions, edited replies and internal notes.
• Technical information such as IP address, browser and device details, where these are actually collected by our hosting provider.
• Support communications you send to us.

Data about social-media commenters may be included when users paste public discussions into the platform. That content is provided by your organisation and processed on its instructions.

Special-category data (such as health, religious or political beliefs, or data about criminal offences) should not be submitted unless it is genuinely necessary, you are authorised to do so, and you have an appropriate lawful condition for it.

• Creating and administering accounts, organisations and team access.
• Providing the analysis, recommendation and human-decision workflow.
• Keeping an auditable record of AI advice and human decisions.
• Securing the service, preventing misuse and investigating incidents.
• Improving reliability and quality of the service.
• Responding to support requests and legal obligations.

• Performance of a contract — creating your account and delivering the service.
• Legitimate interests — security, fraud prevention, service improvement and maintaining organisational audit records. We balance these against individual rights.
• Legal obligation — where we must retain or disclose information by law.
• Consent — used only where genuinely required, and never bundled with service acceptance. You may withdraw consent at any time.

Submitted discussions are analysed by an AI model to suggest categories, scores and draft replies. The AI does not publish anything and does not make final moderation or legal decisions. Threats, safeguarding concerns, criminal allegations, doxxing and live legal matters are marked for mandatory human escalation. AI advises. Humans decide.

PublicVoiceShield sends submitted analysis content directly to OpenAI, which acts as our AI model-processing provider under a data processing agreement (OpenAI Data Processing Addendum v.010126, effective 1 January 2026, executed 6 August 2026). Requests are made server-side from our application; your browser never contacts the model provider. Direct requests are sent with store: false, which disables retention of the response object for account features on supported requests. This is not zero retention: OpenAI may retain API content in abuse-monitoring logs for up to 30 days under its standard API data controls. Zero Data Retention is not enabled.

OpenAI may use sub-processors listed in its published Sub-Processor List under the terms of its DPA. Users must not submit confidential, legally privileged or restricted information unless authorised by their organisation.

Lovable remains our application development and hosting provider. Lovable Cloud (Lovable's managed backend service, built on Supabase infrastructure) provides the database and authentication. Lovable's AI Gateway is retained only as an internal fallback route available to the platform owner; it is not used for live customer analysis. We also use Brave Search for public-source searches you deliberately start, and Zoho Mail for our business mailboxes. Only providers confirmed to be in use are listed on our Subprocessors page, with any unverified detail marked as unconfirmed. Processing locations and international-transfer mechanisms for providers other than OpenAI remain unconfirmed, and are marked as such in each entry.

Our providers may process data outside the United Kingdom. Where that happens, an appropriate transfer mechanism (such as the UK International Data Transfer Addendum) must be in place. Processing locations are being confirmed with each provider and will be documented on the Subprocessors page before public launch.

Retention periods are configurable by organisation and summarised in our Data Retention Policy. Current periods are recommended defaults awaiting administrator and legal approval.

Access is restricted to signed-in users and scoped to their own organisation by database-level access rules. API keys are held as server-side secrets and are never exposed to the browser. Analysis and decision records are append-only.

• Access to your personal data.
• Correction of inaccurate data.
• Deletion, in some circumstances.
• Restriction of processing.
• Objection to processing based on legitimate interests.
• Portability, where applicable.
• Withdrawal of consent, where consent is relied upon.
• Complaint to the UK Information Commissioner's Office (ico.org.uk).

Use the Privacy Rights page, or email privacy@publicvoiceshield.com. We may ask you to verify your identity before acting on a request. Requests are reviewed by a person; nothing is deleted automatically.

We record a version number and a last-updated date for every policy. Where a change is material, signed-in users are shown a one-time notice and asked to acknowledge the new version.