Legal Dictation — Why Privacy Matters for Lawyers
Lawyers have dictated since the profession existed. What changed is where the audio goes. A generation ago, dictation stayed in the room — spoken to a person or a device, both physically present. Today, most dictation tools send your voice to a server in another country, process it on shared infrastructure, and return the text. Somewhere between your mouth and the screen, attorney-client privilege passed through a third party's hands.
This is not a theoretical risk. It is a professional responsibility question that every lawyer using cloud-based dictation should be asking.
Attorney-Client Privilege and Third-Party Servers
Privilege protects confidential communications between attorney and client made for the purpose of legal advice. The protection can be waived — and one of the most common ways to waive it is by voluntarily disclosing privileged information to a third party.
When you dictate a case strategy memo using a cloud transcription service, the audio of that memo is transmitted to the provider's servers. Depending on the service, it may be stored temporarily for processing, logged for quality improvement, or retained for training machine learning models. Even if the provider's privacy policy says otherwise, the data has left your control. You are relying on a third party's promise, not on the architecture of the system.
Bar associations have begun issuing guidance on this. The consensus is clear: lawyers have a duty to understand the technology they use and to take reasonable steps to protect client confidentiality. Using a cloud dictation tool without evaluating its data practices is increasingly difficult to defend as "reasonable."
What Happens When Audio Hits a Third-Party Server
Here is the typical data flow for cloud-based dictation:
- Your microphone captures audio containing privileged information.
- The audio is compressed and sent over the internet to the provider's API.
- The provider's servers — often in a different jurisdiction — process the audio.
- The transcription is returned to your device.
- The audio may or may not be deleted from the server immediately.
At steps 2 through 5, your privileged communication exists outside your control. It crosses network boundaries, sits on third-party hardware, and is subject to the provider's data retention and security practices — practices you likely have not audited.
Wispr Flow, for example, processes audio on cloud servers and costs $180/year. Apple's Enhanced Dictation sends audio to Apple's servers. Even tools that advertise "privacy" often mean "we have a privacy policy," not "your data never leaves your device."
How On-Device Dictation Eliminates the Risk
SpeakUp takes a fundamentally different approach. The entire transcription engine — whisper.cpp, running on Apple's Metal GPU — operates on your Mac. Audio is captured by your microphone, processed by your processor, and the resulting text is typed into whatever application is in focus. No audio is transmitted. No server is contacted. No third party is involved at any point.
This is not privacy by policy. It is privacy by architecture. There is no server to subpoena, no cloud storage to breach, no data retention policy to parse. The privileged communication never exists anywhere except on your machine, exactly as it would if you typed it yourself.
Where Lawyers Use Dictation Most
The daily work of a lawyer is overwhelmingly text. Dictation fits naturally into the tasks that consume the most time:
Case notes and memos. After a client call or court appearance, dictate your notes while the details are fresh. SpeakUp has no time limit — dictate for thirty seconds or thirty minutes.
Briefs and motions. Draft the substance by voice, then edit on screen. Most lawyers find they produce a first draft two to three times faster by dictating than by typing.
Client emails. Especially for lengthy explanations of legal strategy or case status, dictation captures your natural communication style and saves significant time.
Time entries. The most universally dreaded task in legal practice. Dictate your time entries as you finish each task — a few seconds of speech captures what would otherwise be a forgotten billable event at the end of the day.
Contract review annotations. As you read through a contract, dictate your comments and redline notes into a separate document or directly into your review tool.
Cost Comparison
Dragon Legal — the traditional dictation tool for lawyers — costs $500 or more per license, plus annual maintenance fees. Cloud-based alternatives charge $100–$200/year per user, with the audio processing risk described above.
SpeakUp costs €29 once. No subscription, no annual renewal, no per-seat licensing. For a solo practitioner or a small firm, this is the difference between a recurring expense and a one-time purchase that pays for itself in the first week of use.
Getting Started
SpeakUp works in every application where you can type — your document editor, your practice management software, your email client, court filing systems, and browser-based tools. There is nothing to integrate and nothing to configure beyond choosing your hotkey.
For a detailed look at how SpeakUp fits legal workflows, see the Lawyers page. For a deeper explanation of how on-device processing works, see SpeakUp's Privacy Architecture.
Your clients trust you with their most sensitive matters. The tools you use should be worthy of that trust.