Topic · 9 guides
AI and client confidentiality
What happens to client information once it goes into an AI tool: whether it trains a model, how long it is kept, who can read it and when privilege is at risk.
Start here if you are choosing a tool or checking one you already use. Every guide links to the vendor's own terms or the regulator's own text.
Start here
- How to anonymise a document before using AI (step by step for lawyers)
Swapping out names isn't enough. Here's the process that actually holds up, including the step almost every guide skips: metadata and fact-pattern identification.
- What 'do not train on my data' actually means in 2026
A side-by-side of what ChatGPT, Claude, Copilot and Gemini retain when you use the consumer vs. enterprise plans, and where the contractual cliffs are.
- Does ChatGPT Free's opt-out actually stop training on your prompts?
Yes, the toggle works. It just doesn't do what most lawyers assume it does. Here's what OpenAI's own settings actually control, and what they don't.
All confidentiality guides
- 22 September 2026 · 7 min readAre your AI prompts discoverable? What the 2026 rulings actually protect
Two courts said in June that AI chat logs are work product. The protection turns on who typed the prompt, and the exceptions swallow a lot of it.
- 17 September 2026 · 7 min readCan lawyers use AI notetakers? What the NYC Bar's 2026 opinions actually say
Every guide this year says get consent and proceed. The City Bar's newest ethics opinion says the default should be not to record at all.
- 4 August 2026 · 8 min readHow to anonymise a document before using AI (step by step for lawyers)
Swapping out names isn't enough. Here's the process that actually holds up, including the step almost every guide skips: metadata and fact-pattern identification.
- 31 July 2026 · 8 min readDoes using AI waive legal privilege? What United States v. Heppner actually decided
A federal judge ruled a defendant's AI chats weren't privileged. The reason wasn't the AI. It's who directed the use, and that distinction changes what you tell clients.
- 28 July 2026 · 7 min readDoes ChatGPT Free's opt-out actually stop training on your prompts?
Yes, the toggle works. It just doesn't do what most lawyers assume it does. Here's what OpenAI's own settings actually control, and what they don't.
- 24 July 2026 · 8 min readIs Microsoft Copilot safe for law firms?
The AI isn't the risk. Your SharePoint permissions are. Here's what actually determines whether Copilot is safe for client work, and why the free version is a different product entirely.
- 25 June 2026 · 5 min readWhat does the SRA actually say about lawyers using AI?
The SRA hasn't banned AI, and it hasn't blessed it either. Here's what the regulator actually expects from solicitors using AI tools, and the four questions that keep you compliant.
- 24 June 2026 · 6 min readIs it safe to use ChatGPT for legal work in 2026?
ChatGPT Free and Plus train on your inputs by default. Here's what's actually safe for lawyers, what isn't, and how to use AI without breaching client confidentiality.
- 23 June 2026 · 7 min readWhat 'do not train on my data' actually means in 2026
A side-by-side of what ChatGPT, Claude, Copilot and Gemini retain when you use the consumer vs. enterprise plans, and where the contractual cliffs are.
Get the weekly brief.
New guides on confidentiality and the rest of the field, with the primary sources linked.
By subscribing you agree to our Privacy Policy.