FREE · PRIVACY

What's safe to put into AI, and 6 settings to check

Two parts. What to leave out when you're using AI to think through a dispute. Then the settings that decide what your apps keep.

By Abi Odedeyi · CodeFreeIQ

⚖️

In a dispute

Take out the names and the identifying details. Same quality of answer.

📸

Camera roll

The Facebook setting that sends your photos up for suggestions.

🔧

Five more

ChatGPT, Claude, LinkedIn and Gemini, each one a few taps.

Get the list and the settings

Pop your details in and they unlock right here.

No spam. Unsubscribe anytime.

Part 1: in a dispute, what's safe to put in

I'm not a lawyer and this is not legal advice. It's what I do with my own. If the dispute is serious, speak to a solicitor.

SAFE TO PUT IN: the general question ("my supplier delivered late and wants full payment, what are my options"), a timeline with the names taken out, a clause described in your own words, and a letter you're drafting with placeholders like [SUPPLIER] and [DATE].

LEAVE OUT: names, addresses and phone numbers, screenshots of the contract or emails, account numbers, and anything your solicitor has sent you.

Why it matters: what you tell a solicitor can be privileged. What you type into a chat window generally isn't, because an AI tool is not a lawyer. In England and Wales, a "document" for disclosure means anything that records information, so prompts and replies may be disclosable if they're relevant to the case. And deleting a ChatGPT chat removes it from your view straight away, but OpenAI keeps it for up to 30 days, longer if it has to for legal or security reasons.

THE PROMPT I USE: I'm in a disagreement with a [supplier / landlord / former employee]. Here is what happened, with the names taken out: [timeline]. Help me think through my options and what questions I should ask a solicitor. Don't assume anything I haven't told you.

Part 2: the camera roll setting, and five more

1. FACEBOOK · CAMERA ROLL. Facebook app > Settings & Privacy > Settings > Camera roll sharing suggestions. Two switches: suggestions while you browse, and "camera roll cloud processing", which sends your photos up to Meta to make edits and collages. It's opt-in, so check whether you said yes to the pop-up. In the UK since April 2026.

2. CHATGPT · TRAINING. Settings > Data controls > "Improve the model for everyone". Off means your chats aren't used to train the model. It doesn't make it worse at helping you.

3. CHATGPT · MEMORY. Settings > Personalization > Manage memories. Read what it has kept about you and delete what you'd rather it hadn't. Deleting a chat doesn't delete a memory saved from it.

4. CLAUDE · MODEL TRAINING. Settings > Privacy, the model-improvement switch. With it on, Anthropic keeps your chats for up to five years. With it off, 30 days.

5. LINKEDIN · AI TRAINING. Settings & Privacy > Data privacy > Data for Generative AI Improvement. LinkedIn has used UK members' profiles and public posts for AI training since 3 November 2025 unless you switch this off. Private messages aren't included.

6. GEMINI · KEEP ACTIVITY. On gemini.google.com, open Activity and switch off Keep Activity. It's on by default and keeps chats for 18 months. Even with it off, Google holds chats for up to 72 hours to run the service.

Business accounts are different. ChatGPT Business and Enterprise, and Claude's work plans, aren't used for training by default. These switches matter most for anyone on a personal account, which is usually most of a small team.

Sources, checked 29 Sep 2026: Meta newsroom (April 2026, camera roll in the EU and UK), OpenAI Data Controls and Memory FAQs, Anthropic's consumer terms update and Privacy Center, LinkedIn Help (UK data use from 3 November 2025), Google Gemini Help (Keep Activity). Menus move, so if a label looks different in your app, search for it in settings.