Privacy and data

Garden · not yet reviewed Established: widely used

Knowing what information is safe to share with an AI tool, how that tool stores and uses it, and which settings and account types protect you, your organisation and the people you serve.

Also called: data privacy · what's safe to paste · data controls · confidential information · training opt-out

Connections

Understand first

Ideas this one builds on.

• Generative AI (garden)

Related

Often used together or easily confused.

• Agents (garden)
• Human in the loop (garden)
• Memory (garden)
• Onboard AI like a new hire (garden)
• Projects and knowledge (garden)
• Prompt injection (garden)
• Tools and connectors (garden)
• Vibe coding (garden)

Blueprints that practise it

• Hire and onboard an AI copilot for one piece of work (garden)
• Turn messy notes into a status draft you can check (garden)

Resources that teach it

• AI Fluency for Nonprofits (garden)
• How to Improve Interview Feedback Consistency with a Custom GPT (garden)
• OpenClaw: The complete guide to building, training, and living with your personal AI agent (garden)
• Tradclaw: a household starter kit for OpenClaw (garden)
• Securing an Always-on AI Employee (garden)
• An opinionated guide to which AI to use to do stuff (Summer 2026 Edition) (garden)

Competency

• Use responsibly

Find privacy and data on the commons map ↗

You may know it from

• law and human resources confidentiality and need-to-know Share sensitive information only with those who need it for the task, and only as much as they need.
• data protection law data minimisation

In products

The same idea, under different names in different tools.

Provider	What they call it
OpenAI	ChatGPT temporary chats and data controls
Anthropic	Claude incognito chats and privacy settings
Google	Gemini Apps Activity settings

In plain words

Anything you paste, upload or connect goes to the AI company’s systems. Depending on the product, your account type and your settings, it may be stored, reviewed for safety, remembered for later chats, or used to improve future models. Work and enterprise accounts usually have stronger protections than free personal ones.

A few habits cover most situations. Know your organisation’s AI policy and which tools it approves. Don’t paste other people’s personal details, health or financial information, passwords or confidential client material into personal accounts. Remove names and identifying details when you can; the AI rarely needs them to help. Check your settings for training and memory, and use temporary or incognito chats for one-off sensitive questions.

Connectors and agents raise the stakes, because the AI can reach data you didn’t paste in yourself.

Example

A caseworker wants help writing a referral letter. She replaces the client’s name with “Client A”, removes the address and date of birth, gets a strong draft, then adds the real details back in her own document.

How it connects

• Every tool and connector, project and memory feature is a privacy choice.
• Human-in-the-loop review matters most where personal data is involved.
• Apps built by vibe coding that store others’ data need extra care.

Try it

Guess whether your chats are used for training and whether memory is on. Then open your AI tool’s privacy or data settings, check, and change one setting to match what you want.

---

Read this card as plain Markdown: /concepts/con-privacy-and-data.md. Know a parallel from your own field, or a product that implements this? Add it.