Does Wise train on your data?
Wise's assistant terms say your chat interactions are used to "refine the accuracy and improve the output" of an assistant built on third-party language models. Neither document says train.
This clause is real, but it doesn't resolve the training question - see the finding below.
"We use information from your interactions and other data sources to refine the accuracy and improve the output of the Wise Assistant."
- 23 Sept 2026
- 5 Oct 2026
- All customers
- global
- Fetched from the company
- B0ABED8E…6DA0
Why this is still UNCLEAR: Corrected 14 September 2026. The document this entry rested on since 17 August was the wrong page. The help article registered as the privacy policy serves "How is Wise regulated in each country and region?", a list of regulators and licence numbers, so the earlier silence finding was about a page that couldn't have answered. The Personal Customer Privacy Notice itself sits behind an index at wise.com/gb/legal/privacy-policy and renders only in a browser. It was captured that way on 14 September, 293 paragraphs. It contains no "machine learning", no "artificial intelligence", no "model" of the kind this site asks about, and its single "training" is the recording of calls "for training and quality purposes", which is staff training. On 23 September 2026 a second document was read: the Wise Assistant Terms of Use. They say the assistant generates answers from "pre-trained datasets" and the information you provide, and that Wise uses "information from your interactions and other data sources to refine the accuracy and improve the output of the Wise Assistant". That's customer chat data improving an AI system, in Wise's words. What they don't say is whether that means training a model, or whether the language-model providers behind it see your interactions. The way out is not to use it: "using Wise Assistant is completely optional" and you can ask for a person. Transactions and messages outside the assistant are still unaddressed. Wise is being written to.
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.
- Not established
- Not stated
- —
- —
- Not established
How to opt out of Wise AI training
Plan by plan, from what Wise'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 customers
Not graded, so no route is pointed at. Corrected 14 September 2026. The document this entry rested on since 17 August was the wrong page. The help article registered as the privacy policy serves "How is Wise regulated in each country and region?", a list of regulators and licence numbers, so the earlier silence finding was about a page that couldn't have answered. The Personal Customer Privacy Notice itself sits behind an index at wise.com/gb/legal/privacy-policy and renders only in a browser. It was captured that way on 14 September, 293 paragraphs. It contains no "machine learning", no "artificial intelligence", no "model" of the kind this site asks about, and its single "training" is the recording of calls "for training and quality purposes", which is staff training. On 23 September 2026 a second document was read: the Wise Assistant Terms of Use. They say the assistant generates answers from "pre-trained datasets" and the information you provide, and that Wise uses "information from your interactions and other data sources to refine the accuracy and improve the output of the Wise Assistant". That's customer chat data improving an AI system, in Wise's words. What they don't say is whether that means training a model, or whether the language-model providers behind it see your interactions. The way out is not to use it: "using Wise Assistant is completely optional" and you can ask for a person. Transactions and messages outside the assistant are still unaddressed. Wise is being written to.
What this policy has done since we started watching
The record starts here. Wise's 2 documents were first captured on 14 Sept 2026, and there's no earlier copy to compare them against, so this verdict has no history behind it yet. From now on every re-read is diffed against those copies, and anything that moves appears here.
Vendor response
We wrote to Wise on 2 Oct 2026, quoting the clauses above and inviting a correction. Nothing has come back. That's published because a company declining to comment on its own policy is something a reader should know, and because the invitation should be on the record whether or not it's taken up. Any reply goes here, word for word.
Report a wrong clause