Wispr Flow is a popular dictation app, and it is fast. If you are a lawyer, the question to ask of any dictation tool is not how accurate it is. It is where your words go while they are being turned into text. This page lays out what Wispr Flow says about that, what to ask any vendor, and how dictation that runs on your own computer differs.
What Wispr Flow says it does
Wispr describes its app in its own published policies. Two points matter for a law firm:
- Audio is processed on its servers. Its consumer data usage policy says the app records audio when you activate it and sends it to the company's servers to be turned into text.
- It offers controls over what is kept. The company describes a Privacy Mode that does not retain voice or transcription data after processing, and a separate setting to share data to help train its models. Check the current wording, because vendors change their terms.
Why that matters for privileged work
A lawyer's duty of confidentiality covers client information, whatever tool carries it. Sending audio of a client matter to a third party's servers is a disclosure decision. Many firms will decide it is acceptable under a strong contract, and some matters may need the client's consent first. The ABA's Formal Opinion 512 (July 29, 2024) discusses these questions for generative AI tools. Read the opinion. There is a plain-English version in Can lawyers use ChatGPT with client information?
How local dictation differs
The full description is on Private dictation for lawyers.
When Wispr Flow is a reasonable choice
For non-confidential writing, such as marketing copy, personal notes or internal memos with no client information, a cloud dictation app can be a good fit. The point of this page is not that cloud dictation is wrong. It is that the choice should be made on purpose, matter by matter, rather than by whatever is installed on a lawyer's laptop.
Six questions to ask any dictation vendor
- Is the audio sent to your servers? If so, to which providers, and under what contract?
- What do you retain, and for how long? Ask about audio, text and any screen or context data the app reads.
- Is any of it used to train models? Ask whether the default is opt-in or opt-out.
- What does the app read from my screen? Some tools use surrounding text to improve output.
- Can it be limited by matter or by app? Lawyers need to keep one client's terms away from another.
- Can I check the claims myself? A vendor that is confident will welcome a firewall test.
Questions
Is Wispr Flow safe for client information?
That depends on your obligations and its current terms. Its published policy says audio is sent to its servers for processing, and it describes a Privacy Mode that limits retention. For privileged material, read the terms and decide whether that disclosure is acceptable, which may include client consent.
Is there a dictation tool that works without sending audio to the cloud?
Yes. Local dictation runs the speech model on your own computer. VaultScribe is designed this way.
Can I use VaultScribe today?
No. It is a design preview and is accepting founding design partners. It is not available to buy or download.
Read next
Private dictation for lawyersHow the pill, legal vocabulary and polish modes work, all on your own computer.Where VaultScribe fits
VaultScribe is a private AI workspace for law firms, and dictation is one part of it. I am looking for founding design partners, with a non-binding letter of intent and terms agreed on a call.
Wispr Flow is a trademark of its owner. VaultScribe is not affiliated with or endorsed by Wispr. This page quotes the vendor's published policies, which may change, and is not a review of the product. General information, not legal advice.