Privacy and Data Safety — What Never to Put Into AI
text
What data to never put into AI — company policies, personal data, financial info. Practical guidelines for Thai workplace context.
ข้อมูลอะไรไม่ควรใส่เข้า AI เด็ดขาด — ข้อมูลบริษัท ข้อมูลส่วนตัว ข้อมูลการเงิน
Privacy and Data Safety — What Never to Put Into AI
What You'll Learn
What data you should NEVER put into AI tools
How to use AI safely at work
Understanding free vs paid data policies
The Diligence skill from the 4D Framework
The Golden Rule
If you wouldn't post it on a public bulletin board, don't put it in a free AI tool.
Free AI tools may use your conversations to train models. Paid/enterprise tools usually have agreements that protect your data. Always check before you type.
Never Put Into AI (Any Tier)
Customer personal data (names, emails, phone numbers, IDs)
Financial data (account numbers, salary details, budgets with names)
Health information (diagnosis, medications, insurance)
Legal documents under attorney-client privilege
Government/classified information
Credentials (passwords, API keys, access tokens)
Be Careful With (Depends on Tool)
Data Type
Free AI
Paid/Enterprise
Safe Approach
Company strategy
Don't
Check policy
Anonymize first
Source code
Don't (public repos)
Check policy
Use private repos only
Meeting notes with names
Don't
Usually OK
Remove names
Financial reports
Don't
Check policy
Use percentages, not actuals
Client requirements
Don't
Check policy
Genericize details
How to Anonymize Before Using AI
Before pasting work content into AI:
Original: "Our client Siam Commercial Bank wants to migrate
their core banking system from the legacy IBM mainframe
by Q3 2026 with a budget of 50M THB"
Anonymized: "A large Thai bank wants to migrate their core
system from a legacy mainframe by Q3 2026 with an 8-figure
budget. What are the key risks and recommended approach?"
You get the same quality of advice without exposing sensitive data.
Tool-Specific Privacy Settings
Claude (Anthropic)
Free/Pro: Conversations may be used for training. Opt out in Settings → Privacy
Team/Enterprise: Data not used for training. Covered by BAA (Business Associate Agreement)
Projects: Files stay within the project, not used for training
ChatGPT (OpenAI)
Free/Plus: Conversations used for training by default. Opt out in Settings → Data Controls
Team/Enterprise: Data not used for training
Google Gemini
Free: Conversations may be reviewed by human annotators
Workspace (business): Governed by your organization's Google Workspace agreement
AI Use Policy Template for Teams
If your team doesn't have an AI policy, start with these rules:
No customer data in AI tools — ever
No financial data in free AI tools — paid tier with training opt-out only
Anonymize before sharing — remove names, IDs, and specifics
Verify AI output before sharing externally — AI can hallucinate facts and generate plausible but wrong information
Disclose AI usage — if AI helped create a deliverable, note it in your work
Key Takeaway
AI is useful but your data is irreplaceable. When in doubt, anonymize. Use paid tiers for sensitive work. And always remember: AI output needs human review before it reaches clients or stakeholders.
