Does Warp train on your data?
Warp trains and fine-tunes the models behind its AI terminal features on your data. A settings control over AI input collection exists, but is never tied to the training use.
A self-serve opt-out exists
"Develop new products and Service, and improve existing products and features of the Service, including by training, fine-tuning and improving the performance of models that support our AI features;"
- 21 Aug 2026
- 5 Oct 2026
- All plans
- global
- Fetched from the company
- 9051D670…1075
How we read it: Corrected 18 Aug 2026 — this page had said no opt-out was named, which was wrong — and re- sourced 21 Aug to the legal policy Warp asked us to cite; both sentences the grade turns on are verbatim in the old and new captures. Under "AI Inputs and Outputs": "You can control our collection of these inputs and outputs at any time in your settings". It sits in the collection section, never beside the training clause, and never says whether switching it off reaches data already held. Warp's Zero Data Retention agreements bind its LLM providers, not Warp: the clause graded here is Warp training its own models, which no ZDR statement addresses.
opt out, object to, objection, legitimate interest, restrict, withdraw, settings
What the document does offer
- "You can control our collection of these inputs and outputs at any time in your settings" — the control this grade rests on, under "AI Inputs and Outputs"
- "you will have the option not to share terminal data with our servers. Moreover, we will provide tools for deleting and exporting that data" — written in the future tense, as an intention rather than a description
- marketing opt-outs, by instruction in the message or by emailing with "Opt Out" in the subject line
Why none of it applies: Both controls govern COLLECTION; the clause graded here governs USE, and the document never joins them. Asked whether ZDR reaches non-Enterprise plans, Warp sent its docs, which say "ZDR applies across all Warp plans" — then scope agent-request ZDR to "Enterprise teams" further down the same page, while warp.dev/security says it is "available on Warp's Enterprise plan". Three scopes for one commitment. None of them touches the training clause, since all describe what the LLM providers may do.
What Warp said when we showed them this
Corrected us“Warp has Zero Data Retention (ZDR) agreements with LLM providers like Anthropic, OpenAI, and Google, meaning they don't retain or train on any customer AI data. For data collected by Warp, you can opt out of telemetry and data collection anytime by going to Settings > Privacy and toggling off "Help improve Warp" and "Send crash reports." On Business and Enterprise plans, data collection is off by default. [This reply was generated by AI.]”
What each plan can refuse
The same questions, asked of every plan we could identify. Every cell comes from a quoted clause or a documented search - see the evidence above.
- Self-serve toggle
- Not stated
- 52 / 100
- Not stated
- Unknown
How to opt out of Warp AI training
Plan by plan, from what Warp's own policy says about refusing. Where it publishes the steps, they're listed here with the limits they carry. Where there's nothing to switch off, or nothing graded, this says so instead of inventing a menu path.
All plans
Yes. Warp has a self-serve setting.
- Open Settings, then Privacy, or search "privacy" in the Command Palette.
- Toggle off "Help improve Warp".
The switch governs what Warp collects: its docs say that with telemetry off, no console interactions are persisted on Warp's servers.
What this policy has done since we started watching
2 changes detected since 21 Aug 2026. Each one is a diff between two archived copies, and each links to the passage that moved.
Vendor response
Their reply is quoted in full at the top of this page.