Home / Safety / Harvey for law
Harvey AI Corporation (San Francisco, US)Is Harvey safe for law firms?
Harvey publishes limited privacy safeguards. Ask the provider for clarification before you put confidential legal data into it.
Where does Harvey stand in law?
Shared place 30 of 45 options for legal practice (professional tools and general assistants) on the GDPR profile. 6 options, like Harvey, have a limited profile. Each tile is an option we track for this field; click for the check.
Published safeguards
multi-model: OpenAI, Anthropic (via AWS Bedrock) and Google (via Google Vertex) (according to provider); hosted on Microsoft Azure, processing in the EU and Switzerland possible (according to provider); no training on customer data, also contractually excluded with model providers (according to provider); SOC 2 Type II, ISO 27001, ISO 27701 and ISO 42001 (according to provider); data processing addendum
How to use Harvey safely
- Use the business variant and sign a data processing agreement (DPA).
- Do not enter unnecessary personal data into prompts; anonymise where possible.
- Ask the provider explicitly where data is stored and processed.
- Keep a professional in the loop: confidentiality is mandatory.
Stronger or equally strong in legal practice
Legal MindLegal Mind B.V. (Amsterdam, NL)
LegalMikeNovitec B.V. (Groningen, NL)
LexboostLexboost B.V. (Utrecht, NL)
AndriAndri AI B.V. (Amsterdam, NL)
SpellbookRally Legal (Canada)Latest developments
If you choose "Yes, and don't ask again" in Claude Code, the tool saves the command verbatim in .claude/settings.local.json, including passwords or API keys. Apiiro found real secrets in 1,122 of 104,185 public copies on GitHub, including 379 database passwords. Keep the file out of git, use variables for secrets, and replace any keys that have already leaked. source
Via pop-up windows in the HTML preview of code blocks, an attacker could bypass Lockdown Mode and siphon off your chat history, via a manipulated prompt, GPT or shared chat. OpenAI patched the flaw on 18 September 2026; Check Point published the details on 5 November. Users do not need to do anything. source
With message threads in the Claude API, Anthropic stores the conversation, the system prompt and the tools on the server, encrypted in memory and not on disk, until up to one hour after the last use. The feature does not qualify for Zero Data Retention; your other ZDR agreements still apply. If you work with personal data under ZDR, only enable threads if your DPIA covers it. source
Questions
- How safe is Harvey for law?
- Harvey publishes limited privacy safeguards. Ask the provider for clarification before you put confidential legal data into it. Limited or partly unknown safeguards: ask the provider.
- Does Harvey train on my data?
- multi-model: OpenAI, Anthropic (via AWS Bedrock) and Google (via Google Vertex) (according to provider); hosted on Microsoft Azure, processing in the EU and Switzerland possible (according to provider); no training on customer data, also contractually excluded with model providers (according to provider); SOC 2 Type II, ISO 27001, ISO 27701 and ISO 42001 (according to provider); data processing addendum
- What should I watch out for with AI in legal practice?
- Professional secrecy is mandatory. Only enter client data into a business/EU variant with a data processing agreement, where input is not used for training. Always keep a lawyer in the loop.
Source documents
Everything about Harvey Best AI for law firms
Summarised by the editorial team from provider documentation · not legal advice
Last checked: 27 October 2026 · How we check
Source: , checked on 27 October 2026
Every Friday: what changed in AI for law firms
New models, prices, oversight and tools, with sources and what it means for you. One email a week, unsubscribe with one click.





































