Trust center · Last updated April 16, 2026

Trust isn't a marketing line.
It's shipped infrastructure.

SOC 2 Type II, GDPR, CCPA, IAB Tech Lab compliance, ads.txt and sellers.json transparency, brand-safety enforced at the auction layer, and a security disclosure policy you can actually find. Here's what we run, where the receipts are, and who to email if you find a hole.

Certifications

Audited, attested, and renewed on a calendar.

Every certification below is current. Reports and letters are released to qualified buyers under MNDA — request via the trust desk and we'll respond inside one business day.

SOC

SOC 2 Type II

Active · 2026 report

Annual report covering security, availability, and confidentiality. Auditor: Prescient Assurance LLC. Coverage period: Apr 2025 – Mar 2026.

EU

GDPR & UK GDPR

Active

Standard contractual clauses (SCCs), DPA available on request, EU data residency tier offered for Enterprise. Article 28 processor.

CA

CCPA / CPRA

Active

"Service provider" classification. We never "sell" personal information as defined by §1798.140. Opt-out signal honored end-to-end.

IAB

IAB Tech Lab member

Member firm

OpenRTB 2.6, Native 1.2, VAST 4.2, ads.txt, sellers.json, ads.cert 2.0, and TCF 2.2 conformance. TAG-ID certification in progress.

PCI

PCI DSS · SAQ-A

Active

Card payments tokenized via Stripe. We never touch PAN data. Eligible for SAQ-A, the lightest PCI scope.

ISO

ISO 27001

In audit · Q3 2026

Stage 1 audit complete. Stage 2 (certification) scheduled for Q3 2026 with BSI. ISMS scope covers all production systems.

HIPAA

HIPAA

BAA available

For health-tech publishers. We sign BAAs and run an isolated PHI-aware tier with encryption-at-rest keys held in customer KMS.

📜

ads.cert 2.0

Active

Every BBX bid request is cryptographically signed. DSPs can verify supply-chain integrity without trusting the network path.

Brand safety

Brand safety enforced at the auction layer — not in a quarterly report.

Standard OpenRTB bcat (blocked IAB categories) and badv (blocked advertiser domains) filters run inside the eligibility loop, before Benna ever scores a bid. If your block list disqualifies every campaign, you get a clean 204 No Content — never a "best effort" placement.

What's enforced

  • IAB content categories — full IAB Content Taxonomy 3.0 mapping, blocked at bcat per request.
  • Advertiser domain blocksbadv applied per request and per always-on publisher allowlist.
  • Sensitive surfaces — political, alcohol, gambling, weapons, and health-pharma blocked by default; opt-in only with publisher attestation.
  • Generated content gating — ads disabled in conversations flagged by the host app's safety layer; we honor downstream signals like Anthropic's safety_score.
  • Pre-bid creative review — every creative passes human + automated review against our content policy before going live.
  • MFA / clickbait blocking — Made-for-Advertising domain heuristics reject low-quality demand at ingest.
Bid request — brand-safety fields
{
  "id": "bid_req_42a",
  "imp": [{ "id": "1", /* ... */ }],
  "site": { "domain": "cursor.com" },

  // ── enforced inside BBX before scoring ──
  "bcat": [
    "IAB7-39",   // health/pharma
    "IAB8-18",   // alcohol
    "IAB11-4"    // politics
  ],
  "badv": [
    "competitor-a.com",
    "competitor-b.com"
  ],
  "wseat": ["boostboss"],
  "regs": { "ext": { "gdpr": 1 } }
}
i

No-bid is a feature. If every eligible campaign is filtered out by brand safety, the adapter returns HTTP 204 with no body. We never silently relax constraints to fill an impression — that's how brand-safety incidents happen on legacy networks.

Supply chain

Receipts, not promises. Every seller declared.

If you can't trace inventory back to a publisher with a name and a domain, you don't have supply-chain integrity. Our IAB sellers.json is live, our publisher-side ads.txt template is documented, and every BBX bid request carries the standard schain object with complete=1.

ArtifactWhat it isLive URL
sellers.json IAB Tech Lab declaration of every seller_id in our supply chain. DSPs fetch this to verify who they're buying from before bidding. /sellers.json
ads.txt Boost Boss's own ads.txt declaring the resellers authorized to monetize boostboss.ai inventory. Standard IAB format. /ads.txt
app-ads.txt Same declaration scoped for native AI apps that resolve via app store metadata rather than DNS. /app-ads.txt
Publisher template The two lines every Boost Boss publisher should add to their own ads.txt at their domain root. /publisher-ads-txt.txt
schain object OpenRTB SupplyChain object attached to every outbound bid request from BBX. complete=1, single-node default, multi-node for resold inventory. Embedded in every BidRequest
ads.cert 2.0 Every bid request is signed with our private key. Verify against the IAB Tech Lab key registry. Public key: /ads-cert-public-key.pem

For publishers — paste these into your ads.txt

If your AI app monetizes through Boost Boss, mirror these two lines at the root of your domain (e.g. cursor.com/ads.txt). DSPs scan ads.txt before every bid — without this declaration, your inventory may be downgraded or blocked.

your-domain.com/ads.txt
# Boost Boss — authorized seller declaration
boostboss.ai, boostboss-primary, DIRECT, c5e8a91f3e0d2b67
exchange.boostboss.ai, bbx-exchange, RESELLER, c5e8a91f3e0d2b67

# Optional contact line (recommended)
CONTACT=sellers@boostboss.ai
SUBDOMAIN=exchange.boostboss.ai
Data handling

What we collect, what we don't, and where it lives.

Boost Boss is built on MCP signals, not cookies. We don't fingerprint, we don't follow users across sites, and we don't keep raw text. Here's the full inventory.

FieldPurposeRetentionPII?
Anonymous session IDAuction de-duplication and frequency capping within a session.24 hoursNo
Intent tokenVocabulary-matched signal for ranking. Raw text discarded after match.30 days (aggregated)No
MCP tool nameTop-of-funnel ranking signal. Tool name only — no arguments, no payloads.30 days (aggregated)No
Host app domainPersona proxy + publisher reporting.2 yearsNo
Coarse regionRegion-level targeting (us-west, eu-central, apac).30 daysNo
Session length bucket<10m / 10–30m / >30m. Used in ranker.30 daysNo
Auction logsBid request / response / win-loss. Kept for reconciliation, billing, and incident response.13 monthsNo
Advertiser account dataEmail, name, billing address. Not used for ranking — only for the dashboard.Account lifetime + 7 years (tax)Yes

What we never collect. No IP addresses past initial routing, no third-party cookie syncs, no device fingerprints, no MAID, no IDFA, no email hashes, no LiveRamp / UID2.0 / RampID, and no raw prompt text. The Benna whitepaper documents the full signal model — see benna-whitepaper.pdf §5.2.

Sub-processors

We use a small number of vendors for hosting, observability, and billing. Every sub-processor signs a DPA and is subject to the same audit cadence as our internal systems. Material changes are announced 30 days in advance via subprocessors@boostboss.ai.

VendorServiceData categoryRegion
VercelEdge runtime & CDNAnonymous request metadataGlobal (multi-region)
SupabasePostgres & authAccount & campaign dataUS-East / EU-West
CloudflareDDoS & WAFEdge request headersGlobal
StripePaymentsBilling & tax dataUS
DatadogObservability & APMService telemetry (no PII)US
AnthropicCreative review LLMAd copy text onlyUS
AWS S3Long-term log archiveAggregated auction logsus-east-1 (KMS-encrypted)
Security

How the platform is built and broken into.

Encryption everywhere, least-privilege everywhere, audit logs on every privileged action, and a paid bug bounty for anyone who can prove us wrong.

Architecture & controls

  • TLS 1.3 on every endpoint; HSTS preload; mTLS available for Enterprise DSP integrations.
  • AES-256-GCM at rest, KMS-managed keys, customer-managed keys (CMK) on Enterprise tier.
  • Zero-trust internal network — every service-to-service call is mTLS-authenticated and short-TTL token authorized.
  • SAML 2.0 SSO & SCIM for advertiser dashboards on Enterprise.
  • Role-based access control with audit log retention of 13 months; SIEM ingest available.
  • Quarterly external pentests by an independent firm; remediations tracked publicly in the trust portal.
  • Annual red-team exercise covering exchange manipulation, bid stuffing, and creative-injection attack classes.

Vulnerability disclosure

If you've found a security issue — auction manipulation, cross-tenant data leak, RCE in a worker, anything — we want to hear from you, and we pay.

  • Critical — $5,000–$25,000
  • High — $1,500–$5,000
  • Medium — $500–$1,500
  • Low / informational — Hall of fame credit

PGP key, scope, safe-harbor terms, and submission form live at security@boostboss.ai and in our /.well-known/security.txt per RFC 9116.

/.well-known/security.txt
Contact: mailto:security@boostboss.ai
Expires: 2027-04-16T00:00:00.000Z
Encryption: https://boostboss.ai/pgp-key.txt
Acknowledgments: https://boostboss.ai/trust#hall-of-fame
Preferred-Languages: en
Canonical: https://boostboss.ai/.well-known/security.txt
Policy: https://boostboss.ai/trust#disclosure
Documents

Request the paperwork your procurement team needs.

Reports and signed agreements are released under MNDA. Send the request, we'll respond inside one business day with a DocSend or signed PDF.

📄

SOC 2 Type II report

Full Prescient Assurance report covering the 12-month observation window.

View attestation →
📄

Data Processing Agreement

GDPR-compliant DPA with SCCs, Annex II tech & org measures, and EU rep contact.

View DPA →
📄

BAA (HIPAA)

Business Associate Agreement for health-tech publishers. Includes PHI-aware tier addendum.

Request →
📄

Pentest summary

Latest external pentest letter — scope, methodology, findings, remediation status.

Request →
📄

Insurance certificate

$10M cyber liability + $5M E&O coverage. COI issued to your account from our broker.

Request →
📄

Benna whitepaper

Full architecture, training protocol, and A/B evaluation. No NDA required.

Download →
Incident response

What happens when something goes wrong.

Outages, security incidents, and material privacy events all run through the same well-rehearsed playbook. Customers see status, remediation, and a public post-mortem — not silence.

Severity ladder

SevDefinitionCustomer comms
SEV-1Production outage or active security incidentStatus page within 5 min · status@
SEV-2Partial degradation or data integrity riskStatus page within 15 min
SEV-3Single-customer issue or non-critical bugDirect ticket follow-up
SEV-4Cosmetic / observability gapTracked, no comms

Breach notification SLA

If we discover or are notified of a personal-data breach affecting your data:

  • Within 24 hours — initial notification with what we know.
  • Within 72 hours — full disclosure aligned with GDPR Article 33 timelines, including categories of data, rough number of records, and immediate remediation.
  • Within 14 days — public post-mortem with root cause and preventive actions.

Live status: status.boostboss.ai · Subscribe via RSS, Slack, or email.

Talk to the trust team.

We answer in plain English and inside one business day. No vendor questionnaire is too long, no security review is too detailed, no DPA addendum is off the table.

Email trust@boostboss.ai Report a vulnerability