AI Vendor Security & Compliance Brief
ChatGPT
Independent due-diligence summary · every fact links to the vendor's official source
79/100
AI Governance Readiness
Enterprise-Ready
This vendor is rated as Enterprise-Ready with a score of 79 out of 100. Key strengths include confirmed SOC 2 and ISO 27001 certifications, with audit reports available under NDA. However, a material security finding is CVE-2025-43714 (MEDIUM) with no fix listed, and a new sub-processor has been added, meaning a new third party may now process your data. The buyer should request a Data Processing Agreement (DPA) and confirm the remediation timeline for CVE-2025-43714.
Summarized strictly from the source-cited facts below — no outside information. Verify each point against its linked source.
Readiness Breakdown deterministic · evidence-only
- Independent Certification SOC 2 / ISO certifications confirmed via the vendor's trust portal (ISO 27001). Audit report available under NDA — standard enterprise practice.
- Vendor-Stated Compliance Vendor states (cited, not independently audited): BAA Available (HIPAA), GDPR, HIPAA, ISO 27001, SOC 2.
- Customer-Data Training Enterprise terms: does NOT train on customer data (consumer/free tiers may differ — see breakdown).
- Data Processing Agreement No public DPA located — request one during procurement.
- Breach History No known breaches in Have I Been Pwned.
- Vulnerability Exposure 1 known CVE(s); none currently in CISA KEV.
- Email Spoofing Protection (DMARC) DMARC enforced — domain spoofing mitigated.
- Vulnerability Disclosure Policy Publishes a security.txt disclosure policy (RFC 9116).
- Web TLS Certificate Valid TLS certificate in place.
- Legal Transparency 4 legal/policy documents publicly tracked.
Score is normalized over assessed components only — “unknown” items
are shown but never silently counted against the vendor.
Ask This in Your Security Review 1 open items
- Data Processing AgreementRequest the Data Processing Agreement (DPA) and current sub-processor list.
Compliance Posture vendor-stated · cited
| Framework | Status | Source |
|---|---|---|
| GDPR | Stated by vendor | https://openai.com/policies/eu-privacy-policy/ |
| ISO 27001 | Stated by vendor | https://trust.openai.com/ |
| SOC 2 | Stated by vendor | https://trust.openai.com/ |
| HIPAA | Not publicly verified | — |
| BAA Available (HIPAA) | Not publicly verified | — |
As published on the vendor's own trust/compliance pages — not independently audited. Independently verified attestations, when available, appear in the certifications section below. Request the underlying report before relying on these.
Data & Contract Facts deterministic · cited
| Attribute | Value | Source |
|---|---|---|
| Sub-processors (published list) | View document → | https://openai.com/policies/subprocessors |
| Trains on Customer Data key clause |
Free / Pro:
trains on data
ChatGPT Free/Plus: conversations may be used to train models unless you turn off 'Improve the model for everyone' in Data Controls.
cited →
Enterprise:
does not train
ChatGPT Business/Enterprise/Edu and the API: business data (inputs and outputs) is excluded from model training by default; training only occurs if the customer explicitly opts in.
cited →
|
see per-tier citations |
Security Posture authoritative · cited
Known Vulnerabilities (CVE / CISA KEV)
Found
1
Vulnerabilities are usually disclosed after the vendor ships a fix, so most carry a patch. What matters for your risk is whether any are actively exploited (CISA KEV) and whether you run a patched version — patched entries below are a normal sign of an active security-response process, not an open exposure.
Vulnerability Disclosure Policy (security.txt)
Found
1
Email Spoofing Protection (DMARC)
Protected
DMARC enforced and SPF present — spoofing well mitigated.
Web TLS Certificate
Valid
Data Breach History
None found
Queried the authoritative source; no records.
Supply-Chain Security (OpenSSF Scorecard)
Not applicable
Closed-source service — no public source repository; OpenSSF Scorecard (open-source supply-chain) does not apply.
OFAC Sanctions Screening
None found
Queried the authoritative source; no records.
SEC Cyber Incident Disclosures (8-K 1.05)
Not applicable
Privately held — not a US-listed public company, so no SEC 8-K cyber-incident reporting obligation applies.
Security & Compliance Timeline authoritative · dated
Dated, source-cited history from authoritative records (NVD, SEC, CISA KEV). Subscribe to get alerted the moment a new event lands.
- 2025-05-19 CVE CVE-2025-43714 disclosed (MEDIUM · no fix listed)
Certifications Available Under NDA / Trust Center attested · report gated
| Certification | Status | Trust Center |
|---|---|---|
| ISO 27001 | Available via Trust Center | https://trust.openai.com |
An independent audit report exists but is gated behind an NDA or trust-center registration. Request it directly via the vendor's trust center. These count as partial assurance — stronger than a vendor claim, but not an open third-party attestation.
Vendor-Claimed, Not Independently Verified treat as unconfirmed
| CAIQ | claimed_unverified | https://trust.openai.com |
| CYBER INSURANCE | claimed_unverified | https://trust.openai.com |
| FEDRAMP LOW | claimed_unverified | https://trust.openai.com |
| ISO 27017 | claimed_unverified | https://trust.openai.com |
| ISO 27018 | claimed_unverified | https://trust.openai.com |
| ISO 27701 | claimed_unverified | https://trust.openai.com |
| PCI DSS | claimed_unverified | https://trust.openai.com |
| PEN TEST | claimed_unverified | https://trust.openai.com |
| SOC2 TYPE2 | claimed_unverified | https://trust.openai.com |
| SOC3 | claimed_unverified | https://trust.openai.com |
These appear on the vendor's site but we could not confirm an independent audit report. Request the underlying attestation before relying on them.
Tracked Legal & Policy Documents
| Document | URL |
|---|---|
| Ccpa Compliance | https://openai.com/policies/privacy-policy |
| Gdpr Compliance | https://openai.com/policies/data-processing-addendum |
| Trust | https://trust.openai.com/ |
| Vuln Mgmt | https://openai.com/.well-known/security.txt |
How to Obtain Non-Public Documents
These documents were not found at a public URL — which is normal. Many are provided on request, only on enterprise plans, or via the vendor's trust portal. Here is where each lives and what to do to get it.
| Document | Availability | How to obtain |
|---|---|---|
| Data Processing Addendum (DPA) | On request / trust portal | No public DPA link was found. Most vendors provide a DPA on request or let you accept one through their trust/legal portal. Start at the trust center, or email the vendor's privacy team (commonly privacy@<vendor-domain>). Trust center → |
| Sub-processor List | Trust portal / on request | A public sub-processor list was not found. Many vendors publish it behind a trust-portal login or send it on request. Request access through the trust center or from the vendor's privacy/security team. Trust center → |
| Business Associate Agreement (BAA) | On request | OpenAI will sign a BAA for ChatGPT Enterprise and the API where PHI is processed. Request one by emailing baa@openai.com. Trust center → |
| Master Services Agreement (MSA) | Negotiated per contract | The MSA governs enterprise contracts and is negotiated per deal, so there is usually no public link. Self-serve plans are covered by the public Terms of Service instead; for an MSA, ask the vendor's sales team during procurement. Trust center → |
| Service Level Agreement (SLA) | Enterprise tier | A formal uptime/support SLA is generally offered only on enterprise/paid plans and attached to the order form. Ask sales for the SLA exhibit or check the enterprise pricing page; the trust center often summarises uptime commitments. Trust center → |
Continuous Monitoring change-tracking active
1 legal & policy document under change-monitoring since 2026-05-31. 3 tracked changes detected since baseline.
Baa
| Detected | Change | Detail |
|---|---|---|
| 2026-06-15 | Sub-processor Change |
1 new sub-processor(s) added: https://openai.com/policies/subprocessors.
What this means: A new third party may now process your data — check it against your DPA's approved sub-processor list and your notification rights.
|
| 2026-06-08 | Governance Readiness Change |
Governance readiness downgraded: Enterprise-Ready → Conditional. Driven by: 1 new CVE (published from 2025-05-19): CVE-2025-43714. 1 of these have no vendor fix
What this means: This vendor's overall governance posture dropped a tier, driven by 1 change this period (see the summary below) — re-check whether it still meets your bar before renewal or expansion.
|
| 2026-06-08 | CVE / Security Incident |
1 new CVE (published from 2025-05-19): CVE-2025-43714. 1 of these have no vendor fix listed yet (CVE-2025-43714).
What this means: A newly disclosed vulnerability has no vendor fix listed yet — ask for the remediation timeline and confirm your exposure.
|
Ask the Legal Documents grounded · cited
Ask a question about ChatGPT's captured Terms, DPA, Privacy Policy or sub-processor list. Answers are read only from the actual document text and always shown with the exact clause. If the documents don't cover it, we say so — we never guess.
The summary only restates the clauses below it and is verified against them — the verbatim clause is always the source of truth.
Monitor ChatGPT — get alerted when this changes
This brief is a point-in-time snapshot. Vendors quietly revise their DPA, sub-processors, certifications and security posture — and disclose new CVEs. Get a priority email the moment ChatGPT changes something that affects your risk. Built for procurement & security teams.
Free. One email per material change. Unsubscribe anytime. No sales spam.
Every data point above is extracted from the vendor's own official trust, security, or legal
pages and links to its source. This brief contains no scraped sentiment, forum chatter, or
AI-inferred opinion — only verifiable, deterministic facts. Verify each source before
procurement decisions.