Accelerating modern engineering teams

Frequently asked questions

• What Is Warp’s Zero Data Retention Policy?

  Zero Data Retention is available only to our Enterprise plan customers. If you select Zero Data Retention, Warp will implement our Zero Data Retention Standard, meaning that:

  • Warp will not retain or otherwise use for product improvements any text, commands, code, outputs, prompts, or other content submitted, transmitted, or generated by you or any of your authorized users through Warp’s terminal application or AI features.
  • Warp will not retain any of your terminal content or AI inputs/outputs (including conversational data from use of our AI features).
  • Warp will not persistently store, retain, or analyze any of the data you provide through your use of AI features in Warp beyond the immediate execution of the AI requests.

  Warp may, however, collect and retain high-level feature utilization data, including analytics and crash reports, subject to the terms of the Enterprise Plan agreement. Please contact your Warp sales representative to discuss whether you’d like high-level reporting for your account.

  Download Warp Zero Data Retention Note

• How Does Warp’s Zero Data Retention Policy Affect LLM Providers?

  Warp integrates with multiple Large Language Model (LLM) providers to power its AI-driven features. These providers include, but are not limited to:

  • Anthropic
  • OpenAI
  • Fireworks (DeepSeek Model Provider)

  Warp has executed Zero Data Retention agreements with such LLM providers. This means that if you select Zero Data Retention:

  • The LLM providers commit not to train their models on any customer-generated data processed through Warp’s services.
  • The LLM providers commit to delete inputs and outputs following the generation of the relevant output
  • Warp will implement technical and contractual safeguards with LLM providers to execute this commitment.

• What data is sent to LLMs?

  • Agent Mode: Agent Mode is an interactive chat-based mode in Warp that lets you perform any terminal task with natural language. Agent Mode is explicitly invoked by the user and chat, and any supplied context (such as files, user query, or attached command output) is proxied via Warp’s server to the selected LLM provider.
  • Active AI: Active AI proactively sends terminal errors, inputs, and outputs to Warp AI to proactively suggest fixes and next actions. Sensitive information is redacted client side before sending any sensitive terminal content (see below). Active AI can be disabled via an organizational policy.

• How does Warp ensure sensitive data is not sent to the LLM?

  Secret Redaction is automatically applied to any content sent to AI features to prevent any sensitive data being leaked. On the enterprise plan, Secret Regexes can be set for the entire team.

• What code gets sent to AI?

  Agent Mode may request access to code files in response to user queries. Agent mode can be configured to always ask for permission to access files or be enabled for automatic file access depending on your organizational policies.

• How is data sent to the providers?

  All AI requests and responses are currently proxied through Warp’s servers (GCP, US-based), so that prompts and responses can be correctly parsed and constructed.

• Can we use our own LLM?

  Warp supports LLMs hosted by your company, please check with your Warp contact to determine if your hosted model is compliant with Agent Mode API (Anthropic and OpenAI are currently supported). Note that at the moment, AI requests are still proxied through Warp Server. Zero Data Retention Guarantees are automatically enabled when providing your own LLM endpoint.