Privacy Settings Across the Big Three

Every major AI product has a privacy page you've never visited. Here's what to click, toggle, and delete to keep your data yours.

35 min · Reviewed 2026

Your chats are data. Do the settings reflect that?

Free-tier AI products are usually allowed to train on your conversations. Paid tiers usually aren't. But 'usually' isn't a spec — it's a setting buried three screens deep. Here's the map to every privacy control on Claude, ChatGPT, Gemini, and the rest.

The privacy settings that actually matter

Setting	Where to find it	What to pick
Training on your data	Settings → Privacy/Data controls	OFF if sensitive work
Chat history	Settings → Data controls	Keep, but delete periodically
Memory / cross-chat memory	ChatGPT: Personalization \| Claude: Profile	ON if trusted workflow, OFF for shared accounts
Voice recording storage	Advanced Voice settings	Delete on exit
Third-party integrations	Connectors / Plugins	Review quarterly, revoke unused
Model Context Protocol (MCP)	Settings → Beta/Integrations	Audit allowed servers

Claude: privacy settings walkthrough

Claude.ai → Profile icon → Settings → Privacy.
'Help improve Claude' — toggle OFF to opt out of training. Paid tiers (Pro/Max/Team) are already opted out.
'Delete all conversations' — available in the same menu.
Claude for Education: already excluded from training by institution agreement.
API data: by policy, Anthropic does not train on API inputs/outputs.

ChatGPT: privacy settings walkthrough

ChatGPT → Profile → Settings → Data Controls.
'Improve the model for everyone' — toggle OFF to opt out.
'Memory' — ON lets ChatGPT remember facts across chats. Review memories at any time.
'Temporary chat' (top-right switch in a chat) — doesn't save to history, isn't trained on.
Enterprise/Team: SOC 2 Type 2, not trained on by policy.

Gemini: privacy settings walkthrough

gemini.google.com → Profile → Activity.
'Gemini Apps Activity' — OFF to stop Google from retaining your chats.
Even OFF, Google retains conversations for ~72 hours for abuse review.
Workspace users: admin policy may override personal settings.
Google AI Pro/Ultra: per-tenant isolation available for enterprise.

The other tools

Perplexity: Settings → 'AI Data Retention' toggle. Pro users can disable by default.
Grok (on X): privacy controls tied to your X account; disable training under Settings → Grok.
Copilot (Microsoft 365): admin-controlled; personal Copilot has a data controls page.
Notion AI: admin-controlled at workspace level; not trained on your data in Business/Enterprise.
GitHub Copilot: Settings → Copilot; 'Allow GitHub to use my code snippets' toggle.

What you should never paste into any AI, even with opt-outs on

Passwords, API keys, SSH private keys.
Social Security numbers, bank account numbers.
Health diagnoses tied to your name (HIPAA concerns).
Other people's private info without their consent.
Any material under an NDA or trade secret.
Student records covered by FERPA.

A privacy audit checklist

Quarterly AI privacy audit (15 minutes): [ ] Log into each AI tool I've used in the past 90 days. [ ] Confirm training-off for each. [ ] Delete chats older than 30 days that contain sensitive context. [ ] Review Memory entries; delete anything no longer true. [ ] Revoke third-party integrations I haven't used. [ ] Check my email for 'new privacy policy' notices I ignored. [ ] Rotate any API keys I haven't rotated in 6+ months. [ ] Spot-check voice transcripts; delete any containing PII.A once-a-quarter routine that keeps the surface area manageable.

Local models — the privacy extreme

If maximum privacy matters, skip the cloud entirely. Tools like Ollama, LM Studio, and Apple's Foundation Models run LLMs on your own machine. Performance is lower than frontier cloud models, but nothing ever leaves your device.

Your settings tell the company how much of you they can keep. Most people have never checked.
— A privacy engineer

The big idea: defaults are not your friends. Spend 15 minutes per quarter clicking through every AI tool's privacy page. Delete what you don't need, opt out of training, and know what can't be taken back.

End-of-lesson check

8 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-tools-privacy-settings-creators

What is the main idea of "Privacy Settings Across the Big Three"?
1. Every major AI product has a privacy page you've never visited. Here's what to click, toggle, and delete to keep your data yours.
2. Use AI as the final authority for the whole decision
3. Avoid checking the answer once it sounds polished
4. Focus only on speed instead of judgment
Which concept is most central to "Privacy Settings Across the Big Three"?
1. data training
2. privacy
3. Claude privacy
4. ChatGPT memory
Which use of AI fits this topic best?
1. Let the AI decide what matters without your review
2. Use the answer before checking whether it fits the situation
3. Claude.ai → Profile icon → Settings → Privacy.
4. Treat the AI output as automatically correct
What should a careful learner remember about "Opting out doesn't un-train"?
1. Use AI to draft or organize ideas about privacy, then verify before acting.
2. Skip the context so the tool can guess faster
3. Treat the output as private even after sharing it online
4. Use the answer without checking the source
You want to use AI after this lesson. What is the safest next step?
1. Act immediately because the AI answer is written clearly
2. Use AI for drafting and comparison, but verify before publishing or relying on it.
3. Hide uncertainty so the final answer looks cleaner
4. Use private or sensitive details before checking permission
How should AI output about privacy be treated?
1. As proof that no other source is needed
2. As a replacement for context, consent, or expert review
3. As a draft or helper output that still needs human judgment and verification
4. As something that becomes correct when it sounds confident
Name one way to verify an AI answer about privacy.
Which action would help you apply "Privacy Settings Across the Big Three" responsibly?
1. Use the tool to avoid thinking through the tradeoff
2. Keep going even if the output conflicts with a trusted source
3. Treat the AI output as automatically correct
4. 'Help improve Claude' — toggle OFF to opt out of training. Paid tiers (Pro/Max/Team) are already opted out.

← Back to interactive lesson

Tendril · Creators · Tools Literacy

Privacy Settings Across the Big Three

Every major AI product has a privacy page you've never visited. Here's what to click, toggle, and delete to keep your data yours.

35 min · Reviewed 2026

Your chats are data. Do the settings reflect that?

The privacy settings that actually matter

Setting	Where to find it	What to pick
Training on your data	Settings → Privacy/Data controls	OFF if sensitive work
Chat history	Settings → Data controls	Keep, but delete periodically
Memory / cross-chat memory	ChatGPT: Personalization \| Claude: Profile	ON if trusted workflow, OFF for shared accounts
Voice recording storage	Advanced Voice settings	Delete on exit
Third-party integrations	Connectors / Plugins	Review quarterly, revoke unused
Model Context Protocol (MCP)	Settings → Beta/Integrations	Audit allowed servers

Claude: privacy settings walkthrough

Claude.ai → Profile icon → Settings → Privacy.
'Help improve Claude' — toggle OFF to opt out of training. Paid tiers (Pro/Max/Team) are already opted out.
'Delete all conversations' — available in the same menu.
Claude for Education: already excluded from training by institution agreement.
API data: by policy, Anthropic does not train on API inputs/outputs.

ChatGPT: privacy settings walkthrough

ChatGPT → Profile → Settings → Data Controls.
'Improve the model for everyone' — toggle OFF to opt out.
'Memory' — ON lets ChatGPT remember facts across chats. Review memories at any time.
'Temporary chat' (top-right switch in a chat) — doesn't save to history, isn't trained on.
Enterprise/Team: SOC 2 Type 2, not trained on by policy.

Gemini: privacy settings walkthrough

gemini.google.com → Profile → Activity.
'Gemini Apps Activity' — OFF to stop Google from retaining your chats.
Even OFF, Google retains conversations for ~72 hours for abuse review.
Workspace users: admin policy may override personal settings.
Google AI Pro/Ultra: per-tenant isolation available for enterprise.

The other tools

Perplexity: Settings → 'AI Data Retention' toggle. Pro users can disable by default.
Grok (on X): privacy controls tied to your X account; disable training under Settings → Grok.
Copilot (Microsoft 365): admin-controlled; personal Copilot has a data controls page.
Notion AI: admin-controlled at workspace level; not trained on your data in Business/Enterprise.
GitHub Copilot: Settings → Copilot; 'Allow GitHub to use my code snippets' toggle.

What you should never paste into any AI, even with opt-outs on

Passwords, API keys, SSH private keys.
Social Security numbers, bank account numbers.
Health diagnoses tied to your name (HIPAA concerns).
Other people's private info without their consent.
Any material under an NDA or trade secret.
Student records covered by FERPA.

A privacy audit checklist

Quarterly AI privacy audit (15 minutes): [ ] Log into each AI tool I've used in the past 90 days. [ ] Confirm training-off for each. [ ] Delete chats older than 30 days that contain sensitive context. [ ] Review Memory entries; delete anything no longer true. [ ] Revoke third-party integrations I haven't used. [ ] Check my email for 'new privacy policy' notices I ignored. [ ] Rotate any API keys I haven't rotated in 6+ months. [ ] Spot-check voice transcripts; delete any containing PII.A once-a-quarter routine that keeps the surface area manageable.

Local models — the privacy extreme

Your settings tell the company how much of you they can keep. Most people have never checked.
— A privacy engineer

End-of-lesson check

8 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-tools-privacy-settings-creators

What is the main idea of "Privacy Settings Across the Big Three"?
1. Every major AI product has a privacy page you've never visited. Here's what to click, toggle, and delete to keep your data yours.
2. Use AI as the final authority for the whole decision
3. Avoid checking the answer once it sounds polished
4. Focus only on speed instead of judgment
Which concept is most central to "Privacy Settings Across the Big Three"?
1. data training
2. privacy
3. Claude privacy
4. ChatGPT memory
Which use of AI fits this topic best?
1. Let the AI decide what matters without your review
2. Use the answer before checking whether it fits the situation
3. Claude.ai → Profile icon → Settings → Privacy.
4. Treat the AI output as automatically correct
What should a careful learner remember about "Opting out doesn't un-train"?
1. Use AI to draft or organize ideas about privacy, then verify before acting.
2. Skip the context so the tool can guess faster
3. Treat the output as private even after sharing it online
4. Use the answer without checking the source
You want to use AI after this lesson. What is the safest next step?
1. Act immediately because the AI answer is written clearly
2. Use AI for drafting and comparison, but verify before publishing or relying on it.
3. Hide uncertainty so the final answer looks cleaner
4. Use private or sensitive details before checking permission
How should AI output about privacy be treated?
1. As proof that no other source is needed
2. As a replacement for context, consent, or expert review
3. As a draft or helper output that still needs human judgment and verification
4. As something that becomes correct when it sounds confident
Name one way to verify an AI answer about privacy.
Which action would help you apply "Privacy Settings Across the Big Three" responsibly?
1. Use the tool to avoid thinking through the tradeoff
2. Keep going even if the output conflicts with a trusted source
3. Treat the AI output as automatically correct
4. 'Help improve Claude' — toggle OFF to opt out of training. Paid tiers (Pro/Max/Team) are already opted out.

← Back to interactive lesson