Why does personal AI use need a hygiene checklist?

Consumer AI tools often retain chat history and may use inputs to improve models by default. Without deliberate settings and habits, personal details, work secrets, and other people's data can be stored or used in ways you did not intend. A checklist turns vague caution into concrete steps.

What should never go into a consumer AI tool?

Government IDs, full payment card numbers, passwords and recovery codes, health records, other people's personal data without their knowledge, and anything covered by an employer NDA or confidentiality duty. If a leak would be a real problem, keep it out of consumer tools.

Does turning off chat history protect me?

It helps — disabling history and model-training toggles reduces retention and reuse of your inputs — but it is not absolute. Providers may still keep data briefly for abuse monitoring, and settings can reset after updates, so combine it with minimising what you share.

Why isolate AI accounts from my main login?

Using a dedicated email and a unique strong password for AI tools limits the blast radius if that account is breached and keeps AI activity separate from your primary identity. It also makes it easier to delete everything later by abandoning one isolated account.

Is my input stored by this tool?

No. The checklist is generated entirely in your browser from your selections. Nothing you choose is uploaded, logged, or saved, so you can answer honestly without any privacy cost.

Personal AI Safety & Privacy Hygiene Guide

A privacy checklist tuned to how you use AI

Consumer AI tools are convenient and leaky by default — many retain your chats and may train on your inputs unless you change settings. This guide asks which tools you use, what you use them for, and how sensitive your inputs are, then produces a personalised hygiene checklist: which settings to turn off, how to isolate your AI accounts, prompt habits to adopt, and the categories of data that should never enter any AI tool. It all runs in your browser.

How it works

You select your tools (ChatGPT, Claude, Gemini, Copilot, and others), pick your main use cases, and set a sensitivity level. The tool maps each choice to concrete actions. Selecting a tool that trains on inputs by default surfaces the exact opt-out toggle to find; a high sensitivity level adds stricter prompt hygiene and a longer “never share” list; work-related use adds employer and NDA warnings. The result is an ordered checklist you can work through once and re-run whenever your tool mix changes.

Tips and notes

Settings drift. Re-check training and history toggles after major app updates — they sometimes reset.
Isolate the account. A dedicated email plus a unique password limits the damage from any single breach.
Redact before you paste. Swap real names, IDs, and numbers for placeholders when the answer does not need them.
Assume retention. Treat anything you type as potentially stored, and keep the truly sensitive material out entirely.