--- id: public-ai.security-and-privacy locale: en visibility: PUBLIC status: LIVE last_reviewed: 2026-08-02 owner: Influblog Security and Privacy --- # Security and privacy Influblog separates public product information from internal company knowledge, tenant data, and executable operations. ## Availability **Status:** LIVE as a platform boundary. Future AI interfaces remain subject to the same role, tenant, consent, and audit requirements as human-facing interfaces. ## Role and tenant boundaries Brands, creators, agencies, and administrators use different roles and product surfaces. Access is checked on the server. Brand Website scope separates website-specific campaigns, products, integrations, affiliate activity, and related reporting inside a brand company. Agency access depends on an explicit managed-brand or represented-creator relationship. Administrator access is operational and audited; it is not a public data source. ## Data categories Public documentation describes product concepts. It does not contain customer campaigns, creator contact details, private messages, individual contracts, wallet balances, unpublished analytics, credentials, or internal operational records. Media and documents follow the same relationship and tenant rules as their database records. Possessing a storage identifier or guessed URL is not authorization. ## Consent and integrations Social and commerce connections require the relevant account authorization and provider permissions. Creator analytics also depends on creator consent. Integration credentials remain server-side and are not exposed through public documentation. ## AI and agents Public AI discovery remains read-only product documentation. Separately, Influblog offers a controlled BETA MCP interface to pre-registered OAuth-PKCE clients for Brand, Creator, and Agency users. Every grant is bound to the signed-in person, exact role surface, tenant, approved scopes, and—where applicable—the selected Brand Website. The allowlist exposes validated reads, expiring safe drafts and two BETA Brand campaign-write requests. A safe draft cannot save, publish, message, order, pay, accept a contract, change an integration, or delete a business object. The person can review, adopt, reject, and disconnect a client from the corresponding Influblog role app. Revocation, rate limits, and audit remain server-enforced. Influblog does not currently publish a general customer OpenAPI, anonymous MCP access, dynamic client registration, or service-account credentials. The Brand write requests can save one AI concept as a private campaign draft or request publication of one complete campaign draft. They require an exact immutable effect preview, idempotency, current authorization and a separate 15-minute explicit confirmation by the same signed-in Brand user unless the user instead granted the dedicated, revocable campaign-execution scope. Generic OAuth consent or adoption of a safe draft does not authorize either action. For paid publication, funds already received by Influblog must be reserved in the Brand wallet before public visibility. A separate wallet-spend scope can use only available wallet funds; missing funds return a secure top-up action. The agent cannot access card data, bypass payment-provider verification, or confirm a user-action-required wallet top-up. See the public [Privacy policy](https://influblog.com/privacy) and [Terms](https://influblog.com/terms) for the current public policies.