Healthcare and dental practices, law firms, accounting firms, financial services, and professional services offices. To be precise about what this is: these controls can support an organization's broader security, privacy, and compliance program. No technology by itself makes a practice HIPAA compliant or satisfies a specific regulation.
AI Governance
Your employees are going to use AI. The question is whether you control how.
ChatGPT, Microsoft Copilot, Claude, Gemini, Perplexity — they are already open on the screens in your office. The decision in front of you is not whether AI gets used. It is which tools are approved, who can use them, and what company information is allowed to go into them. We put those boundaries in place and keep them working.
We'll review how AI is being used in your business today.
01 The questions to answer first
02 The false choice
You don't have to choose between blocking AI and ignoring the risk.
Businesses usually land on one of two extremes, and both create problems. There is a third position, and it is the one we implement.
03 The third option
Blocked, wide open, or governed.
Most offices end up with one of the first two by default. This is what governed use looks like day to day: which tools are approved, what information is not allowed to leave, and how you would know if it did.
04 Which tools
Decide which AI tools belong in your business.
Instead of every employee choosing an application on their own, your business runs an approved AI environment. We don't control the AI vendors — we control access to them, and what your employees can do inside them, according to your policy.
How a request flows
The controls behind that flow
| Control | What it looks like in your office |
|---|---|
| Approved services | The AI tools the business has reviewed stay available to the people who need them |
| Unapproved services | Applications outside the approved list are blocked, with a path to request review |
| By department and role | Marketing, accounting, clinical, and administrative staff can each have different rules |
| Corporate accounts | Where it matters, employees must sign in with the company account rather than a personal one |
| Discovery | AI applications already in use — including ones management doesn't know about — get identified |
| New tools | As new AI services appear, policy is extended rather than rewritten |
05 Which information
Stop sensitive information before it leaves.
An AI tool is a new path out of your business. The employee using it usually has good intentions — they are trying to finish a task faster — and still paste something into a prompt that should never have left the office. Security controls can recognize sensitive information and stop an employee from pasting or uploading it into an AI service when doing so would violate company policy. Security professionals call this data loss prevention, or DLP.
Allowing a tool is not the same as giving it your information
What happens at the moment of submission
06 In practice
Policy is granular, not all-or-nothing.
Four situations we've built policy around. In each one, the useful AI work continues and the specific risk is removed.
07 Visibility
Know how AI is being used across your business.
This is not monitoring what employees type. It is the record of which services are in use and which policies were enforced — the same kind of reporting you already get on backups, patching, and MFA coverage.
| What's reported | Why it matters to you |
|---|---|
| AI services in use across the business | Shows what is actually being relied on before you write policy about it |
| Which groups or departments are using them | Tells you where guidance and training need to go first |
| Attempts to send protected information | Documents that the control worked, and where the pressure is |
| Policy violations and blocked activity | Evidence for insurers, clients, and your own internal review |
| Usage patterns over time | Signals when a rule is too tight, too loose, or out of date |
08 Three parts, not one
AI governance is not a piece of software.
A published AI acceptable use policy with nothing enforcing it is a document nobody follows. Technical controls with no policy behind them produce confusing blocks and help tickets. The combination is what works.
09 How we run it
Six steps, then ongoing review.
AI platforms change monthly and so does the way your staff use them. Governance is a standing item, not a one-time configuration.
10 Who needs this most
Offices that hold other people's information.
Where this lives
On top of the security baseline
11 Start here if
Find out how AI is really being used in your office.
A free 90-minute review of AI use, exposure, and policy.