A team code for AI at work has five points: tools, data, disclosure, review and escalation. One named person owns it. A person checks every AI draft before it goes to a manager or a client. This post gives you the one-page code.
Learn these skills in the course AI at Work (BL-AIW), module 3.
Who This Is For
Managers and HR staff who see their teams use AI with company data.
The Five-Point Code
| Point | The Team Rule | Example |
|---|---|---|
| 1. Tools | Use only the AI tools the company approves. | The company Team or Enterprise plan, not a free personal account |
| 2. Data | Do not paste confidential data into a tool that is not approved. | No customer lists, price lists or staff records |
| 3. Disclosure | Say when AI wrote a large part of a document. | A note in the email or the file name |
| 4. Review | A person checks the facts, numbers and tone before it goes out. | The human in the loop accepts, rewrites or rejects the draft |
| 5. Escalation | Report a data leak or a wrong output at once. | Tell the code owner the same day |
Free consumer AI tools can train their public models on your uploads. Team or Enterprise plans keep the data separate.
The Do-Not-Paste List
- Customer names with phone numbers, emails or IC numbers.
- Price lists, margins and quotations not yet sent.
- Staff salaries, appraisals and medical records.
- Contracts, legal letters and board papers.
- Passwords, keys and system access details.
The Check Before Any AI Draft Goes Out
- Is every fact, name and number correct? Check them against the source.
- Does the draft promise anything we cannot deliver?
- Does it sound like us, in our tone?
- Is any confidential data inside it?
- Would I sign this with my own name?
Common Mistakes to Avoid
- Sending an AI draft without reading it.
- Pasting a customer sheet into a free chatbot “just to sort it”.
- A code with no owner. Name one person.
Try It This Week
Write the five points on one page for your team. Name the owner. Ask each person to read and sign it.
Frequently Asked Questions
Which AI tools does the course use?
The course practises with Claude, Copilot and ChatGPT. The method works with other tools too.
What is the human in the loop?
The person who checks AI output and decides to accept, rewrite or reject it.
Next Step
Read the full outline of AI at Work (BL-AIW). Public class: 14 and 15 December 2026. All dates are on the Training Calendar. For a group from your company, get a quote for in-house training. More Work Tips are in the skill area Personal Effectiveness. Read also:
