Home › Is Claude safe for confidential info?
Is Claude safe for confidential info?
Short answer: it depends on your plan and settings. For consumer accounts (Free, Pro, Max, including Claude Code on those accounts) there is a “Help improve our AI models” setting; when it is on, your new chats and coding sessions can be used for model training. Below is what Anthropic's own pages say, checked Oct 23, 2026. This is a summary, not legal advice.
How to stop Claude training on your chats
- Desktop or browser: your name → Settings → Privacy → turn off “Help improve our AI models”.
- Mobile: your name → Settings → Privacy → turn it off.
What to know before you paste anything
- Turning it off stops new chats and coding sessions being used for future training; data already in a training run or a trained model stays there.
- Chats flagged by safety classifiers may still be used for trust-and-safety work.
- Anthropic's Claude Code data-usage docs give retention for consumer accounts: 5 years if you allow model improvement, 30 days if you don't.
Work and business accounts
The same docs say commercial users (Team, Enterprise and API) have a standard 30-day retention period, and Claude's pricing page lists model training as “None by default” on Team and Enterprise.
Is it safe for confidential information? A checklist
- Remove names, account numbers, health details and anything covered by a confidentiality agreement before you paste.
- If the information belongs to your employer or a client, use the tool your organisation has approved, on a business plan, signed in with your work account.
- Turn off model training on personal accounts, and remember that turning it off doesn't delete past chats.
- Don't paste passwords, API keys or private keys into any chatbot.
- Check your industry rules (health, legal, finance, government) before using any consumer AI tool for work.
When it fits / when it doesn't
Fits: public information, your own drafts, and anonymised notes, with training turned off.
Doesn't fit: client or employer data your organisation hasn't approved for this tool, regulated data (health, financial, legal privilege), passwords and keys.
Sources (official pages, checked Oct 23, 2026)
Compare every tool's privacy setting on one page: Is my data safe? · Claude features and prices · Compare
Other tools: ChatGPT · Gemini · Microsoft Copilot · Grok · DeepSeek. Vendors change these pages often; check the source before relying on a detail.
New: what Claude saves when it joins a call: Claude Meetings privacy.