ClauseLens

Does Harvey Train on Your Data? Plan-by-Plan Answer with Quotes

Harvey states that it does not use customer inputs, outputs, or uploaded documents to train underlying models, and that a bespoke model is created only if a customer explicitly requests one.

Last checked Nov 16, 20267 verified quotes6 source documentsRead by Claude Haiku 5.5 · quotes verified by code

Trains on your data

Does not train

DPA

Yes

Data location

US by default; EU/Switzerland or Australia localization available

Retention

Customer-configurable; options as short as three hours

Plan by plan

“All plans” means the vendor’s documents do not distinguish plans for this question.

Trains on your data

  • EnterpriseDoes not train

How to opt out

  • EnterpriseNot stated

Data retention

  • EnterpriseCustomer-configurable; options as short as three hours

Where data is stored

  • EnterpriseUS by default; EU/Switzerland or Australia localization available

DPA available

  • EnterpriseYes

Subprocessors

  • EnterpriseMicrosoft, OpenAI, Google Cloud, AWS, Anthropic, Mistral, ElevenLabs, DeepL, and others

Certifications claimed

  • EnterpriseSOC 2 Type II, ISO 27001, ISO 27701, ISO 42001, AIUC-1

Human review of your data

  • EnterpriseNo human review

What Harvey says, word for word

Open a field to read the vendor’s sentence, its source and our confidence.

Trains on your dataAll plansDoes not train
“Harvey contractually guarantees through our Platform Agreement that your data stays yours. We don’t use inputs, outputs, or uploaded documents to train underlying models.”

Contractual guarantee via Platform Agreement

How to opt outAll plansNot stated

The vendor’s documents we read do not say this. We show “Not stated” instead of guessing.

Confidence: n/aChecked Nov 16, 2026
Data retentionAll plansCustomer-configurable; options as short as three hours
“Customers define their own data retention periods, with options as short as three hours, allowing them to tailor data handling to their specific risk profiles and internal governance requirements.”

Model providers must use Zero Data Retention

Where data is storedAll plansUS by default; EU/Switzerland or Australia localization available
“Harvey hosts its cloud environment in Microsoft Azure. For customers with requirements or preferences for data localization we offer processing in the EU and Switzerland or Australia.”

Localized processing for customers with requirements

DPA availableAll plansYes
“This DPA is part of Your Terms with Harvey.”

https://www.harvey.ai/legal/data-processing-addendum

DPA: Data Processing Addendum | Harvey ↗Confidence: highChecked Nov 16, 2026
SubprocessorsAll plansMicrosoft, OpenAI, Google Cloud, AWS, Anthropic, Mistral, ElevenLabs, DeepL, and others
“If you selected a Harvey processing location that isn't listed for Anthropic or Mistral on our subprocessor list, you can still use their models through AWS Bedrock and Google Vertex.”

Full list at harvey.ai subprocessor page

Subprocessors: Subprocessor List | Harvey ↗Confidence: highChecked Nov 16, 2026
Certifications claimedAll plansSOC 2 Type II, ISO 27001, ISO 27701, ISO 42001, AIUC-1
“Harvey undergoes annual SOC 2 Type II and ISO 27001 audits to validate these controls.”
Human review of your dataAll plansNo human review
“No human eyes on customer data: We call this “eyes off,” and it means that no human has access to or reviews our customer data.”

Customer data is 'eyes off'

Change history

No changes detected since we started tracking Harvey on Nov 16, 2026 (version 1). We re-check its documents regularly.

Sources we read

More on Harvey

Other ai productivity tools

AI Productivity Tools that don’t train on your data →

Informational, not legal advice. Every verdict is what the vendor’s own documents state, quoted word for word. Read the cited source before you decide, and check your own contract: negotiated terms can differ.