Data Governance · 4 min read

Only 14% of MSPs Have Full Visibility Into Customer AI Tools

Only 14% of MSPs have full visibility into customer AI tools, while 56% of paying Copilot tenants lack both DLP policies and sensitivity labels. A 406-MSP survey finds demand outruns readiness.

By Priya Raman726 words

Key takeaways

  1. Inforcer surveyed 406 MSPs globally; 14% have full visibility into customer AI tools and 63% have partial visibility.
  2. 76% of MSPs said customers ask about AI at least monthly; 47% field questions weekly.
  3. 56% of customer tenants paying for Copilot lack both DLP policies and sensitivity labels; 16% have both.
  4. 62% of MSPs offer Copilot readiness services, versus 46% offering Purview or data protection.
  5. 41% of MSPs cite data hygiene as the most common obstacle to Copilot deployment.
MSPs On Customers Rushing To Deploy AI: ‘It’s Like A Kid In A Candy Store’ - CRN
MSPs On Customers Rushing To Deploy AI: ‘It’s Like A Kid In A Candy Store’ - CRN — imgix / unsplash

Only 14% of managed service providers (MSPs) say they have full visibility into the artificial intelligence (AI) tools their customers use. Inforcer’s MSP Microsoft Readiness Report surveyed 406 MSPs globally: 76% said customers ask about AI at least monthly, 47% field questions weekly, and 63% have only partial visibility.

Why are readiness conversations becoming harder?

Michael Goldstein, market president of Southeast Florida at Fort Myers, Fla.-based Entech, described the sales challenge to CRN. “They are very anxious about getting that AI out there. When we talk about the readiness assessment, it’s a hard sale. They just want it. They don’t care about the ramifications. It’s like a kid in a candy store.”

Goldstein puts the core problem in management, not merely in technical safeguards installed by an MSP. Executives need to understand what the organization is deploying and its exposure; technical teams can roll out the technology, but management must understand the risks.

What is already inside customer environments?

At Austin, Texas-based GSC Technologies, cloud security analyst Brenda Scott begins each engagement by identifying what employees already use. In many tenants, staff have adopted several AI tools before the business decides on policy, including websites, desktop apps, Office add-ins and browser apps.

GSC then catalogs which platforms employees access, how they use them and what information they enter. Scott’s central concern is confidential material—client data, financials or internal documents—being pasted into a tool the company does not control. Tool count therefore matters less than the data entering each tool.

Governance cannot be a one-time conversation. AI tools change, new forms appear beyond those blocked a year earlier, and more employees use them each month. GSC consequently revisits the issue with clients as usage grows, making governance a recurring operating process rather than a launch checklist.

Where do customers bypass MSP advice?

Stanley Louissaint, founder and president of Watchung, N.J.-based MSP Fluid Designs, says clients increasingly arrive with implementation decisions rather than requests for advice. His team often hears from them only after permission problems or other technical roadblocks appear.

Fluid Designs does not refuse deployment. It explains risks, recommends appropriate safeguards and leaves the final decision to the customer. Louissaint also cites the ease of employee self-service: a person can sign up for ChatGPT in all of 30 seconds and connect a Google account, with no technical barrier necessarily stopping that action.

What does Copilot expose?

Inforcer co-founder Will Connor said the survey’s surprise was how quickly MSPs began delivering AI without first establishing governance. “People are already delivering AI, but they haven’t caught up,” he said. “They’ve got the AI in first before they’ve really got their hands around the data.”

The gap often begins before AI deployment. MSPs have often migrated files into SharePoint without addressing permissions and access controls. Copilot can then surface information employees cannot access, including a payroll run or redundancy list that was never properly permissioned or labeled.

Inforcer found that 56% of customer tenants paying for Copilot have neither data loss prevention (DLP) policies nor sensitivity labels enabled. Only 16% have both protections. The service mix reinforces the mismatch: 62% of MSPs offer Copilot readiness services, compared with 46% offering Purview or data protection services.

Another 41% of MSPs identified data hygiene, meaning the condition of underlying data, as the most commonly cited obstacle to Copilot deployment. The figures suggest that Copilot readiness is being sold faster than the controls required to support it.

What should data teams do next?

Data teams should first inventory AI use across websites, desktop apps, Office add-ins and browser apps. They should record which platforms employees access, how they use them and what information they submit, then identify confidential inputs moving into tools the company does not control.

Before Copilot rollout, teams should repair permissions and access controls, enable DLP policies and sensitivity labels, and address data hygiene. They should also schedule recurring governance reviews because tools, access methods and user counts change; one assessment cannot govern a changing environment.

Executive owners should understand exposure before deployment, while technical teams explain risks and safeguards. Finally, MSPs should conduct an AI readiness exercise before selling AI licensing. This sequence treats readiness as a data-control prerequisite rather than a paperwork exercise around a purchase.

Google News: Data Governance (Source)

Price tags

Share this article:

More from Priya Raman

Priya Raman

Show full bio

Correspondent covering regulation, policy and their impact on Analytics Digest is a professional daily at Analytics Digest.

13 articles

Read also

Next article »