Data Privacy Checklist

Protect Client Information When Using AI Tools

AI tools process the text you give them. Before pasting anything, ask: "Would I be comfortable if this showed up in the output or training data?"

Before You Paste: The 3-Second Check

STOP and ask:

Is there a name? Remove or replace with "Client A"
Is there an address? Remove or use "123 Main St" placeholder
Is there financial info? Remove specific numbers
Is there personal info? (SSN, DOB, account numbers) Never paste these

If you answered YES to any: sanitize first, then paste.

Quick Sanitization Guide

Original	Sanitized Version
John and Sarah Smith	"my clients" or "the buyers"
742 Evergreen Terrace	"[property address]"
Pre-approved for $650,000	"approved for [amount]"
SSN: 123-45-6789	NEVER include
Bank account info	NEVER include
Birth dates	NEVER include
Email addresses	Remove or "[client email]"
Phone numbers	Remove or "[phone]"

What's Safe to Share with AI

Generally safe:

Property features (beds, baths, square feet)
Neighborhood names (public info)
General market conditions
Transaction stages ("under contract," "inspection complete")
Generic scenarios ("a first-time buyer who...")
Your Context Card info
Template prompts with placeholders

Safe with caution:

Listing descriptions (public anyway)
General feedback patterns ("showings say price is high")
Transaction timelines
Negotiation strategies (keep vague)

Never share:

Full names paired with addresses
Financial documents
Pre-approval letters
Tax returns
Social Security numbers
Bank/account information
Medical information
Legal documents in full
Passwords or login credentials

AI Tool Privacy Tiers

Not all AI tools are equal. Know what you're using:

Tier 1: Consumer Tools (Most Common)

ChatGPT (free/Plus): Data may be used for training (opt-out available)
Claude (free): Data may be used for training
Most free tools: Assume data is not private

Best practice: Sanitize everything. Don't share sensitive info.

Tier 2: Business/Enterprise Tools

ChatGPT Enterprise/Team: Data not used for training
Claude Pro with workspace: Better privacy controls
Microsoft Copilot (business): Enterprise data protection

Best practice: Still sanitize, but more protected.

Tier 3: Offline/Local Tools

Local LLMs (like running Ollama on your computer)
Air-gapped systems

Best practice: Data never leaves your device. Safest option.

Chat History Settings

In ChatGPT:

Settings → Data Controls
Toggle OFF "Chat history & training"
Conversations won't be used for training
Note: May limit some features

In Claude:

Check workspace/account settings
Review data usage policies
Use Projects for better organization

Regular cleanup:

Delete sensitive conversation threads
Don't let conversations accumulate indefinitely
Start fresh chats for different clients/topics

Deal Sheet Privacy

When using Deal Sheets:

DO include:

Property address (needed for context)
General buyer/seller situation ("nervous sellers," "first-time buyers")
Transaction status
General goals

DON'T include:

Full legal names
Contact information
Financial specifics beyond price
Personal medical/family details
Anything you wouldn't say to a colleague in public

Team AI Policies

If you have team members using AI:

Establish:

Which AI tools are approved
What information can/can't be shared
Sanitization requirements
Who reviews AI outputs
Data retention/deletion policies
Training on privacy practices

Document and enforce consistently.

State Privacy Laws

California (CCPA/CPRA):

Stricter consumer data protections
Right to deletion
Disclosure requirements

Other states with privacy laws:

Virginia, Colorado, Connecticut, Utah (as of 2024)
More states adding regulations

Check your state's requirements and adjust practices accordingly.

Client Consent Considerations

When to consider getting consent:

Processing significant personal data through AI
Using AI for analysis that affects decisions
When clients explicitly ask about your tools

Sample consent language:

Our team uses AI-powered tools to improve service efficiency.
Information you share may be processed through these tools.
Your data is [describe protections]. You may opt out of AI
processing by letting us know your preference.

Incident Response

If you accidentally shared sensitive data:

Immediately: Delete the conversation/chat thread
Assess: What was shared? Who could be affected?
Document: Record what happened and when
Notify: Broker and possibly affected clients (consult legal)
Prevent: Update your process to prevent recurrence

Better to over-disclose than under-disclose if there's a breach.

Weekly Privacy Audit

Every Friday (5 minutes):

Review AI chat history — delete sensitive threads
Check that new prompts follow sanitization rules
Verify team members are following protocols
Update Context Card if you notice new data types appearing

The Golden Rule

Treat AI chat boxes like public whiteboards.

Would you write this information on a whiteboard in a coffee shop?

No → Don't paste it into AI
Yes → Safe to use

From the AI Acceleration Resource Room