Brave Leo — privacy & data policy
Provider: Brave. PromptPrivacy score: 88/100. Last verified October 18, 2026.
What happens to your data?
When you use Brave Leo, your queries and relevant webpage context are sent to the backend server stripped of identifiers like your IP address. The servers process your request ephemerally and do not retain prompts, responses, or use them for model training. Your conversation history stays stored locally on your device unless you choose to sync or submit feedback.
Does it train on your data?
Brave does not use user prompts, context, or conversations for model training.
Can you opt out?
Yes. Model training is disabled by default without needing an opt-out. Users can opt out of saving conversation history by turning it off in the browser settings.
Data retention
Brave Leo does not store or retain prompts, responses, context, or personal data on its servers. Any large cached prompts used for performance are deleted within minutes. Conversation history is kept locally on the user's device unless deleted by the user. If a user voluntarily submits feedback via thumbs-up or thumbs-down, the rating and conversation text are retained on servers for up to 1 year.
Encryption & security
Conversation history is stored locally on the user's device. If Brave Sync is enabled, conversation history is included in Sync chains which use client-side encryption where only the user holds the decryption key. Specific in-transit encryption protocols and security certifications (e.g., SOC 2, ISO 27001) for the AI backend servers are not specified in the policy.
Is it safe for work?
Yes, Brave Leo provides strong privacy protections for workplace use because it does not store prompts on servers, strips IP addresses, and does not train AI models on user data. However, organizations that require centralized administrative controls, compliance auditing, or formal Enterprise Data Processing Agreements (DPAs) should note these are not outlined in the policy.
Source policy: https://brave.com/privacy/browser/