Your Confidential Information Never Goes Into Public AI. Full Stop.
As your Employer of Record, we hold some of the most sensitive information that exists about your business and your people.
Salaries. Social Insurance Numbers. Banking details for payroll. Employment contracts. Health information tied to benefits. The commercial terms of our agreement with you. If that information were mishandled, the damage would land on your employees and your business.
So before I tell you how we use AI, let me tell you the rule that governs all of it.
The Rule
We never enter confidential client data into public AI systems.
Not client names. Not employee names. Not salaries, SINs, banking details, contract terms, or anything else you have trusted us with.
Not “just this once to save time.” Never.
When my team uses AI for research, drafting, or analysis, they work with public information and anonymised, generalised scenarios.
“What are the termination notice requirements for a five-year employee in British Columbia?” is a question we might explore with AI support, and then verify against the legislation.
“Draft a termination letter for your employee’s actual name” is not, and never will be.
Why This Rule Exists
Public AI tools are extraordinary, and that is exactly why they need boundaries.
Information entered into a public AI system leaves your control. Depending on the tool and its settings, it may be retained, reviewed, or used in ways neither you nor we can see or undo. For the categories of data an EOR handles, that is simply not an acceptable risk, whatever the productivity upside.
There are also legal obligations. Canadian privacy law, including PIPEDA and its provincial counterparts, places real duties on how personal information is collected, used, and safeguarded. Our clients often have their own obligations on top. A provider that is casual about where AI tools sit in that picture is a provider creating risk for you.
And honestly, there is a simpler reason. We built this company on trust and personal service. It is fundamental to who we are. You cannot claim to offer either while feeding client information into systems you do not control.
What Responsible Use Looks Like In Practice
The rule is the headline. Underneath it sit working safeguards:
- Clear Internal Boundaries – Everyone on our team knows what can and cannot go near an AI tool. It is not left to individual judgment in the moment. It is policy, and it is trained.
- The Right Tool For The Right Data – Public AI tools are for public information. Client information lives in the secure, access-controlled systems we use to run payroll and manage accounts, the same standard of care we have always applied, with AI changing nothing about where sensitive data is allowed to live.
- Anonymise Before You Analyse – When AI helps us think through a client scenario, the scenario is stripped of anything identifying first. The pattern goes in, your details do not.
- Human Review On Everything – Every AI assisted output is reviewed by an experienced member of our team before it reaches you, which is also our last line of defence against anything inappropriate ever slipping through.
Ask Your Provider This Question
If you are evaluating any EOR, PEO, or Payroll Provider in 2026, ask them directly: “What is your policy on entering client data into AI tools?”
A good provider will have a clear, immediate answer. A worrying one will improvise. You have just read ours, in writing, from myself, the Managing Director.
Your information is safe with us. That was true before AI, and we have made sure it stays true now.
Martin Balcome
Managing Director, Bridgewater Talent Integration
