// Trust & Security Report

    Anthropic PBC logo

    Anthropic PBC

    Claude (consumer chat, Claude Code, Team/Enterprise, Anthropic API/Platform, Claude for Government)

    Certifications held

    10

    Maturity

    Enterprise

    Trains on your data

    No

    Trust center

    Yes

    // Certification ledger

    Each held certification is backed by a verbatim quote from the vendor's own trust or security page. “Not confirmed” means we could not verify it publicly, not that the vendor lacks it.

    SOC 2 Type 2
    HELD

    Scope | SOC 2 Type 2 ... Claude via Anthropic's API ✅ | Claude Enterprise ✅ | Claude Team ✅. Resources: [Anthropic] 2025 Type 2 SOC 2 and CSA STAR L2 Report.pdf

    Verify on trust.anthropic.com
    SOC 3
    HELD

    Resources > SOC 2: [Anthropic] 2025 Type 2 SOC 3 Report.pdf

    Verify on trust.anthropic.com
    ISO/IEC 27001:2022
    HELD

    Compliance: ISO 27001:2022. Resources > ISO: [Anthropic] ISO 27001 Certificate (2025).pdf

    Verify on trust.anthropic.com
    ISO/IEC 42001:2023 (AI management system)
    HELD

    Compliance: ISO/IEC 42001:2023. Resources > ISO: [Anthropic] ISO 42001 Certificate (2025).pdf

    Verify on trust.anthropic.com
    CSA STAR Level 2
    HELD

    Scope ... CSA Star: Claude via Anthropic's API ✅, Claude Enterprise ✅, Claude Team ✅. Report: [Anthropic] 2025 Type 2 SOC 2 and CSA STAR L2 Report.pdf

    Verify on trust.anthropic.com
    HIPAA (BAA available)
    HELD

    Compliance: HIPAA. Resources > HIPAA: [Anthropic] 2025 Type 1 HIPAA - Report (C4E).pdf, [Anthropic] 2025 Type 1 HIPAA Report (1P API).pdf. FAQ: Is Anthropic willing to sign a BAA?

    Verify on trust.anthropic.com
    NIST 800-171
    HELD

    Scope ... NIST 800-171: Claude via Anthropic's API ✅, Claude Enterprise ✅, Claude Team ✅. Resource: [Anthropic] 2026 NIST 800-171r3 Attestation Letter

    Verify on trust.anthropic.com
    FedRAMP High
    HELD

    Compliance: FedRAMP High. Claude for Government ... FedRAMP High ✅. Resource: Claude for Government (C4G) PFCS-SS FedRAMP High Authorization Package

    Verify on trust.anthropic.com
    DoD IL4 / IL5
    HELD

    Claude on Amazon Bedrock in AWS GovCloud ... DoD IL4 ✅ | DoD IL5 ✅ (Only applicable to the partner hosting environment where Claude models are running)

    Verify on trust.anthropic.com
    UK Cyber Essentials
    HELD

    Compliance: UK Cyber Essentials. Resource: [Anthropic Ireland Limited] Cyber Essentials Certificate (2025)

    Verify on trust.anthropic.com

    // Privacy & AI training

    Trains on customer data

    No

    Data processing agreement

    Offered

    Data region

    Hosted on Google Cloud, AWS, and Microsoft Azure (worldwide); Cloudflare for CDN. Regional/data-residency options available via Amazon Bedrock, Google Vertex AI, Microsoft Foundry, and Claude for Government (US GovCloud).

    Split by product. COMMERCIAL (Anthropic API, Claude for Work/Team/Enterprise, Claude Gov, Education): no training by default - 'By default, we will not use your inputs or outputs from our commercial products (e.g. Claude for Work, Anthropic API, Claude Gov, etc.) to train our models.' Training only happens if the customer explicitly opts in via thumbs up/down feedback. CONSUMER (Free, Pro, Max, and Claude Code on consumer accounts): as of the Aug 2025 consumer terms update, chats and coding sessions ARE used for model training UNLESS the user opts out in Privacy Settings; opting in extends retention to 5 years, opting out keeps the 30-day retention. Incognito chats are never used for training.

    // Security controls

    Bug bounty / responsible disclosure

    HackerOne - public Bug Bounty Program (infrastructure/code, CVSS-based rewards) plus a Vulnerability Disclosure Program (anthropic-vdp) and a separate Model Safety Bug Bounty. Public Responsible Disclosure Policy with 3-business-day acknowledgement.

    anthropic.com

    Penetration testing

    Annual third-party penetration testing; report available in Trust Center ([Anthropic] 2025 Annual Penetration Testing Reports).

    trust.anthropic.com

    Encryption

    Customer-managed encryption keys (CMEK) offering documented; encryption design whitepaper published in Trust Center.

    trust.anthropic.com

    Subprocessors transparency

    Published subprocessor list (GCP, AWS, Azure, Cloudflare, TurboPuffer) with update notices.

    trust.anthropic.com

    Security advisories / CVE handling

    Publishes security advisories (e.g. CVE-2026-22561 DLL search-order hijacking in the Windows installer).

    trust.anthropic.com

    Accessibility (VPAT/ACR)

    Maintains current VPAT/Accessibility Conformance Reports for Claude.ai Enterprise web, iOS, and Android.

    trust.anthropic.com

    // Products & data scope

    Claude (Free / Pro / Max - consumer)Consumer AI assistant

    data: Personal chats/coding sessions. Used for model training by default UNLESS opted out (post-Aug 2025 consumer terms). 5-year retention if opted in, 30-day if opted out. No SOC2/ISO scope claim at consumer tier; covered under Anthropic's overall compliance posture but not the enterprise contractual protections.

    Pricing: Free $0; Pro ~$17/mo annual; Max from $100/mo. Lowest data-protection tier - reviewers should treat consumer Claude as train-by-default-unless-opted-out.

    Claude TeamSMB/team collaboration

    data: No training on customer data by default. In scope for SOC 2 Type 2, ISO 27001, ISO 42001, CSA STAR, NIST 800-171. HIPAA listed N/A for Team.

    Mid tier between consumer and Enterprise.

    Claude Enterprise / Claude for WorkEnterprise AI assistant

    data: No training on customer data by default. Full cert scope: SOC 2 Type 2, ISO 27001, ISO 42001, CSA STAR, HIPAA, NIST 800-171. DPA and BAA available. SSO supported.

    Strongest contractual data protections; recommended tier for regulated buyers.

    Anthropic API / Claude PlatformDeveloper API

    data: No training on inputs/outputs by default. In scope for SOC 2 Type 2, ISO 27001, ISO 42001, CSA STAR, HIPAA, NIST 800-171. Also available via AWS Bedrock, Google Vertex AI, Microsoft Foundry (partner-managed compliance).

    Data residency achievable through cloud partner regions.

    Claude for GovernmentPublic sector / regulated gov

    data: FedRAMP High authorized; DoD IL4/IL5 via AWS GovCloud Bedrock. No training on customer data.

    Distinct compliance package for US government workloads.

    Claude CodeAgentic coding tool

    data: Inherits the plan it runs under: train-by-default-unless-opted-out on consumer accounts; no-training-by-default on commercial/enterprise accounts.

    Data scope depends entirely on the underlying account tier.

    // What to watch

    • Training behavior differs by product: commercial/enterprise/API = no training by default; consumer Free/Pro/Max = trains by default unless the user opts out (Aug 2025 consumer terms change). Display this distinction clearly so buyers do not assume consumer Claude is as protected as Enterprise.
    • Self-hosting not available; data residency only via cloud-partner regions.
    • Some certs (DoD IL4/IL5, HIPAA on Bedrock/Vertex) are partner-managed in the cloud hosting environment, not standalone Anthropic certifications.

    // At a glance

    Pricing model

    Freemium + subscription (Free / Pro ~$17/mo / Max from $100/mo) + usage-based API + custom Enterprise/Gov contracts

    Self-hostable

    No

    // How we verified this

    Every certification marked HELD is confirmed against a verbatim quote on Anthropic PBC's own trust, security, or privacy pages. We reject certifications claimed only on third-party aggregators, on a cloud host's behalf, or by a similarly named company.

    Last verified 2026-06-27. Compliance changes over time. Always confirm directly with the vendor before relying on any certification for a purchasing or compliance decision.

    trust.anthropic.com

    > Browse all vendor trust reports