B2B Privacy Policy (draft)
Status: DRAFT — pending counsel review (Linear AIP-9).
Not for publication or customer use. Dev-only artifact for counsel markup (Linear AIP-121; owner answer 7, 3 Nov 2026).
Product scope: self-serve B2B API and related operator surfaces; United States and Israel only at launch.
1. Who this policy is for
This Privacy Policy describes how AI Powered Social Scan (“we”, “us”, “our”) handles personal data when we provide our business-to-business (B2B) social-media screening platform to customer organizations (“Customers”, “you” in the organizational sense).
It applies to:
- Representatives of Customer organizations who sign up, manage accounts, or use our API, developer portal, or (when enabled) customer portal; and
- Individuals whose public social-media content Customers ask us to collect and analyze (“Scan subjects”, “data subjects” in the GDPR sense where applicable).
This draft assumes Customers use the service for lawful business purposes (for example employment, vendor, or compliance screening) and remain responsible for their own legal bases and notices to Scan subjects, as described below.
2. Our role: controller vs processor
| Data category | Typical role of AI Powered Social Scan | Typical role of Customer |
|---|---|---|
| Customer account and company data (signup questionnaire, billing contacts, API keys, usage, support) | Data controller — we decide how we operate the platform account | Customer is the organization identified on the account; its representatives act on its behalf |
| Scan subject data (public profile metadata, public posts, derived scores, reports, cached media tied to a scan) | Data processor (or equivalent service provider) — we process on documented instructions from the Customer via API/portal actions | Data controller (or equivalent) — Customer chooses who to scan, receives results, and determines purpose |
Counsel note: Exact controller/processor (and joint-controller) characterizations under US state privacy laws, Israeli PPA, and GDPR-style frameworks must be confirmed per jurisdiction and contract stack (AIP-9).
We do not use Scan subject data to build advertising profiles or to sell personal information. Operational telemetry, security logs, and billing metadata are handled separately from report content.
3. Categories of personal data we process
3.1 Customer company and account data (“Customer data”)
Collected at signup and during account use, including:
| Source | Examples (non-exhaustive) |
|---|---|
Access questionnaire (POST /v1/access/applications) | Company legal name, website, country of registration, contact name, role, email, declared use cases, expected scan targets and subject countries, consent/informed flags, expected volume |
| Account operation | Tenant identifier, API key identifiers (secrets stored as hashes only), credit ledger events, webhook configuration, daily limits, operator notes |
| Billing (when enabled) | Payment method tokens via payment processor; invoicing metadata — not full card numbers on our systems |
| Support and compliance | Operator review status (green / yellow / red access doors), re-review flags tied to rules-matrix version, audit notes |
Automatic signup runs a rules engine (counsel-configured matrix) that assigns a green, yellow, or red door before scans are unlocked; yellow/red paths may require operator approval (AIP-91).
3.2 Scan subject data (“Screening data”)
Processed only when a Customer submits a scan job. Content is drawn from publicly available social profiles and posts (and related public media), then analyzed by our pipeline (scrape → transcribe/media → rank → judge → finalize). Examples:
- Identifiers and profile fields visible on the network (names, handles, profile URLs, avatars where public)
- Public post text, timestamps, engagement metadata, and public audio/video we transcribe or archive for evidence
- Derived outputs: risk flags, grades, word-frequency portraits, PDF/JSON reports, token-gated report links, in-report disputes
We may cache stage outputs per tenant/profile to reduce cost and latency; cache entries carry TTL-style expiry in infrastructure.
Not in B2B reports (product decision): certain B2C-only elements such as Snitch payloads and sentiment/topic color overlays (AIP-276).
3.3 Data we do not intentionally collect
Government ID numbers, health records, or non-public sources outside the Customer-directed scan workflow — unless they appear in public posts the Customer’s job targets (in which case they are processed only as part of Screening data).
4. Purposes of processing
| Purpose | Data | Legal basis (draft — counsel to finalize) |
|---|---|---|
| Provide the B2B platform (signup, auth, scans, reports, webhooks) | Customer data; Screening data | Contract; legitimate interests of Customer and us |
| Access eligibility and fraud/abuse prevention | Customer data; questionnaire evaluation | Legitimate interests; legal obligation where applicable |
| Billing and credits | Customer data; usage | Contract |
| Quality assurance on outputs | Screening data (sampled) | Legitimate interests; contract |
| Security, audit, and incident response | All categories | Legitimate interests; legal obligation |
| Backup, disaster recovery, and retention | All categories | Legitimate interests; legal obligation |
Operational QA (owner defaults, dev configuration): manual QA on the first three scans for each new Customer, then 10% spot checks, with processing region aligned to the Customer’s company country (US or Israel only at launch). Counsel should confirm whether QA sampling requires additional notice or DPA terms.
5. Geographic scope and data location
Customers: We currently accept self-serve signup only for companies registered in the United States or Israel. The access questionnaire captures country_of_registration and subject-country declarations; the rules engine enforces launch policy.
Infrastructure: Production workloads are designed around AWS eu-central-1 (Frankfurt) as primary region, with cross-region replica and backup patterns documented for disaster recovery (AIP-124 / AIP-127). This draft does not yet state final “data residency” marketing claims — counsel must align customer-facing language with DPA and transfer tools.
Transfers: Where Screening data or Customer data is accessed from the US or Israel, or stored/processed in the EU, we rely on appropriate transfer mechanisms (counsel to specify: SCCs, Israeli adequacy/regulations, US state addenda, etc.).
6. Retention
Retention varies by category. The table below reflects platform design intent and backup matrices; counsel must reconcile with statutory minima/maxima and customer contracts.
| Category | Retention (draft intent) | Notes |
|---|---|---|
| Customer account & questionnaire | Life of account + limited post-termination period | Counsel to set termination/audit window |
| API keys | Until revoked; hash retained for abuse forensics if counsel approves | Plaintext secret shown once at creation |
| Scan jobs, reports, artifacts | Customer-configurable minimums not yet public — default platform retention TBD with counsel | Backup: in-account ~35d; cross-account DR vault ~90d; S3 lifecycle tags (e.g. 90d / 365d replica) per AIP-124 |
| Stage cache | TTL-based expiry on cache index | Reduces re-fetch of public content |
| Webhook delivery logs | ~30 days (operational) | Align with OpenAPI/delivery store behavior |
| Budget/COGS operator events | ~90 days | Admin audit trail |
| CEO/read-only analytics exports | Up to 1 year for aggregated reports; 90d tag on some objects | See internal ops docs — not customer-facing |
Erasure: We support tenant-scoped deletion workflows for dev/operator tooling; production data subject erasure paths (especially for Scan subjects who contact us directly) must be finalized with counsel and wired to customer instructions (AIP-127 erasure safety).
7. Sharing and subprocessors
We use subprocessors to run the service, including cloud infrastructure (AWS), payment processing (Stripe when enabled), email delivery, and data vendors that retrieve public social content (Bright Data is prioritized in vendor chains per platform policy). A public subprocessor list and notice process are counsel deliverables (AIP-9).
We do not sell Customer or Screening personal data. We may disclose information if required by law, to protect rights and safety, or in connection with a merger/acquisition subject to counsel review.
8. Security
We apply industry-standard measures appropriate to the sensitivity of Screening data: encryption in transit and at rest (KMS-managed keys), tenant isolation in the data plane, hashed API secrets, access controls on operator tools, and monitoring. No security program is perfect; we maintain incident response procedures documented in docs/ops/incident-response.md.
Details suitable for customer security questionnaires should be split into a Security Appendix (counsel + security review).
9. Data subject rights and requests
Rights depend on jurisdiction. This section describes operational intent pending counsel approval.
9.1 Customer representatives (account data)
Individuals acting for a Customer may contact us to access, correct, or delete account/profile data tied to their organization, subject to identity verification. Some requests may require Customer admin authorization.
9.2 Scan subjects (screening data)
Because Customers are controllers for Screening data:
- Primary path: Scan subjects should direct access, correction, deletion, restriction, or objection requests to the Customer that ran the scan.
- Direct to us: If a Scan subject contacts us, we will acknowledge and, unless legally prohibited, forward or facilitate the request with the relevant Customer, or respond when we act as controller (for example pure account data).
We will maintain a privacy@ (or equivalent) inbox and internal routing to operator/compliance roles — address TBD at publication (counsel question).
Timelines: Statutory deadlines (CCPA/CPRA, Israeli PPA, etc.) to be mapped in the final policy and DPA.
9.3 Appeals and complaints
We will describe appeal paths and relevant supervisory authorities in the final US/Israel versions.
10. Automated decision-making
The service uses automated scoring (LLM-assisted rank/judge stages) to prioritize public posts for human review in reports. Customers should not rely solely on automated scores without their own review. Counsel should assess whether additional GDPR Article 22 / US state “profiling” disclosures apply.
11. Children
The B2B service is not directed to children. Customers must not submit scans targeting minors except where permitted by law and their own policies; counsel to confirm age thresholds (13/16/18) per jurisdiction.
12. Open questions for counsel
- Controller/processor labels — Confirm for US state laws, Israel PPA, and any EU touch via Frankfurt hosting + EU subprocessors.
- Legal bases table — Finalize lawful basis per purpose (contract vs legitimate interest vs consent) for Customer data and for our minimal first-party signup processing.
- Customer vs platform notices — What must appear in Customer’s privacy notice to Scan subjects vs this B2B policy vs a DPA?
- DPA template — Standard processing terms, audit rights, subprocessors, breach notice SLAs, and instructions for erasure/deletion.
- Cross-border transfers — SCCs, UK IDTA, Israeli regulations, and US state transfer disclosures for EU-hosted processing serving US/IL Customers.
- Retention defaults — Contractual retention for reports/artifacts; interaction with backup/DR copies (AIP-124); zero-retention options for LLM vendors where available.
- Direct Scan subject requests — When may we delete without Customer consent? How to handle multi-tenant cache keys?
- Sensitive/inferred data — Treatment of potentially sensitive content in public posts; restrictions on use for “high-risk” categories.
- QA sampling — Legal basis and transparency for first-three-scans and 10% spot-check human review.
- Marketing communications — Opt-in/out for product email to account contacts (distinct from transactional mail).
- Publication gate — Alignment with AIP-9 counsel checkpoint before any customer-facing link goes live.
- Terms cross-links — Relationship to Terms of Service, Acceptable Use, and access questionnaire representations (
subject_consent,subject_informed).
13. Changes to this policy
We may update this policy after counsel review and with notice to Customers as required by law. Material changes to processing will be reflected in DPAs and/or in-product notice.
14. Contact
Privacy inquiries (draft placeholder): privacy@aipoweredsocialscan.com — confirm domain and mailbox with counsel.
Postal / registered agent addresses: TBD for US and Israel entities.
Appendix A — AIP-121 / owner answer 7 coverage map
| Requirement (AIP-121 / task scope) | Section(s) in this draft |
|---|---|
| Customer company data we hold | §3.1, §4 |
| Scanned people / public post data | §3.2, §4 |
| Our role for each category | §2 |
| Retention | §6 |
| US and Israel scope | §5, §3.1 (signup) |
| Data subject requests | §9 |
| DRAFT + AIP-9 counsel gate | Header, §12 Q11 |
| Open questions for counsel | §12 |
Document version: draft 2026-11-03 — repo path docs/legal/drafts/b2b-privacy-policy.md.