// Trust & Security Report
H2O.ai
Enterprise AI platform for predictive and generative AI with on-premise, hybrid, and managed cloud deployment options. Core products: h2oGPTe (generative AI), H2O Driverless AI (AutoML), H2O LLM Studio (fine-tuning), tabH2O (foundation model), plus open-source H2O-3, Wave, MLOps, Feature Store.
Certifications held
3
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.
Verify on trust.h2o.ai“SOC 2 Type 2 - Listed as current certification in Trust Center; HAIC Security Guarantees Model confirms 'SOC 2 Type 2 Certification: Demonstrates effective security controls operating over time'”
Verify on trust.h2o.ai“H2O.ai achieves SOC2 Type 2 + HIPAA/HITECH Report with Unqualified Status”
Verify on trust.h2o.ai“IRAP - Listed as current certification in Trust Center”
> Show 4 unconfirmed / not-held certifications
source: h2o.aiFedRAMP 'In Process' designation at High Impact Level as of May 14, 2025. NOT full certification/authorization yet. Press release: 'H2O.ai Achieves FedRAMP "In Process" Designation at High Impact Level' - actively undergoing FedRAMP authorization process, not yet authorized.
source: trust.h2o.aiCompany is pursuing ISO 27001 but has not yet achieved certification. Search results consistently indicate 'pursuing ISO 27001' with no evidence of current certification.
source: trust.h2o.aiNo public evidence of ISO 42001 certification. Not listed in trust center or security documentation.
source: trust.h2o.aiNo public evidence of CSA STAR or CSA STAR for AI certification.
// Privacy & AI training
Trains on customer data
No
Data processing agreement
Offered
Data region
Primary processing in US; customers can select data residency options where available. International transfers use EU-U.S. Data Privacy Framework and Standard Contractual Clauses (Module 2).
Explicit prohibition on customer data use for training. Privacy Policy states: 'Customer-provided input data, metadata, and model data are not used to train H2O.ai AI models, unless explicitly agreed in writing.' DPA further states: 'H2O.ai will not use Customer Data to train, retrain, fine-tune, or improve any artificial intelligence or machine learning models, except solely for Customer's benefit and only where expressly agreed in writing.' No use for advertising, profiling, or cross-customer model training allowed.
// Security controls
Encryption at rest
Industry-standard algorithms for data at rest; encrypted storage for sensitive data
docs.h2o.aiAccess Control
Role-based access control (RBAC) following least privilege principles; Identity integration via Keycloak OpenID Connect (OIDC) for external identity provider integration
docs.h2o.aiAudit & Logging
Comprehensive audit logging of user activities and access patterns; logs retained up to 3 years
docs.h2o.aiVulnerability Management
Code reviews, dependency scanning, license verification, vulnerability scanning, annual penetration testing
docs.h2o.aiBackup & Disaster Recovery
Automated backups using Velero for Kubernetes resources; access-controlled backup environments; customer-specific RTO/RPO objectives
docs.h2o.aiData Deletion
Upon termination, personal data deleted or returned within 30 days at customer request, unless legally required to retain
h2o.aiSubprocessor Management
Maintained subprocessor list at trust.h2o.ai; customers can object within 30 days on reasonable data protection grounds; unresolved objections allow service termination without penalty
h2o.ai// Products & data scope
Data it handles: Customer data processed on-premise, hybrid, or managed cloud per deployment choice
Enterprise agentic AI platform; achieves 75% accuracy on GAIA benchmark; used by federal government (NIH case study)
Data it handles: Multi-model LLM support; customer data not used for training without consent
Secure, enterprise platform for SLMs and LLMs; air-gapped deployment capable; used by AT&T (2X ROI case study)
Data it handles: Automatic feature engineering on customer data
Model development acceleration; explainability included; used by Commonwealth Bank Australia (70% fraud reduction)
Data it handles: Tabular data only; no model training required; no data stored
CSV-to-predictions foundation model; zero data residency
Data it handles: Private data fine-tuning; no training on public data without consent
SLM distillation and fine-tuning on private data
Data it handles: On-premise only; 2M+ open-source users
Open-source AutoML; Python, R, Spark support
Data it handles: Fully managed or hybrid cloud; on-premise deployment options
Unified platform across cloud, on-premise, and hybrid environments; FedRAMP High In Process authorization
// What to watch
- Homepage marketing states 'Airgapped and FedRAMP,' but current status is 'In Process' at the High level (achieved May 2025), not full authorization.
- Trust center FedRAMP listing shows 'FedRAMP Certified Class D,' but the latest press release (May 2025) shows 'In Process' at the High level. This may reflect a discrepancy between products or an outdated trust center listing.
- ISO 27001 remains 'pursuing' and is not yet certified. Marketing language describing a certification as 'in pursuit' means it has not yet been achieved.
- AI governance certifications (ISO 42001, CSA STAR) not evident in public documentation. As AI governance standards mature, H2O may pursue these in future.
// At a glance
Pricing model
Subscription-based with managed cloud and hybrid/on-premise options; details require direct contact
Self-hostable
Yes
// How we verified this
Every certification marked HELD is confirmed against a verbatim quote on H2O.ai'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 2026-07-07. Compliance changes over time. Always confirm directly with the vendor before relying on any certification for a purchasing or compliance decision.
trust.h2o.ai