A man reviews a cybersecurity dashboard with compliance indicators, analytics, AI tools, and cloud services displayed.
Businesses are adopting AI faster than they can put in the security controls needed to protect sensitive data. That is the headline claim of new research from inforcer, a vendor of Microsoft 365 management tools for MSPs. The numbers are worth reading closely, and so are the caveats that come with them. The problem is less that Copilot is broken and more that the surrounding tenant hygiene is patchy.

A man reviews a cybersecurity dashboard with compliance indicators, analytics, AI tools, and cloud services displayed. What inforcer actually measured​

The MSP Microsoft Readiness Report 2026 combines two datasets. They are different populations and shouldn't be mixed up.

  • Survey: 406 MSPs managing Microsoft 365 environments across EMEA, North America, and Australia and New Zealand.
  • Platform data: anonymized, aggregated data from inforcer's multi-tenant platform, covering 64,709 client tenants managed by 1,525 MSPs across the three regions.

Inforcer is a vendor that sells tenant-management and Copilot-readiness tooling. The data comes from its own customer base and survey audience, so it isn't necessarily representative of all Microsoft 365 tenants or MSPs. The figures are also not independently audited.

The headline numbers​

Copilot demand is outrunning the foundations

  • 82% of surveyed MSPs are already selling or piloting Microsoft 365 Copilot.
  • Most MSPs help with identity (81%) and devices (80%). Only 46% say they offer data protection services to govern the data feeding AI applications.

Tenant-level controls

  • Platform data shows 56% of Copilot tenants lack a DLP policy, sensitivity labels, or both. Only 16% have both.
  • The reporting doesn't say how that 56% divides between missing DLP, missing labels, or neither. "56% have no safeguards at all" would be a misreading.
  • 86% of MSPs lack full visibility into customers' AI usage.

Consistency and monitoring

  • 69% say they have centrally managed or framework-aligned best practices, but only 28% apply them consistently.
  • 51% lack a consistently enforced baseline across customer tenants.
  • Just 18% continuously review customers' Microsoft security posture. About 32% review once a year or only after an incident.

Automation

  • 80% call automation a top priority for the year ahead.
  • Only 31% can push a single policy change across all managed tenants within minutes.
  • Inforcer's landing page adds that 41% automate less than 10% of routine tenant changes. That comes from the same report, so it isn't independent corroboration.

The landing page also says 14% of MSPs have full visibility of the AI tools their customers use. That looks like the flip side of the 86% figure, but the question wording and denominators aren't confirmed, so treat the two as related rather than identical.

Why labels and DLP matter for Copilot​

A careful reading matters here. The report doesn't show that Copilot bypasses permissions. Microsoft's documentation says Copilot respects existing access controls. Microsoft's own admin guidance puts it this way: Copilot respects existing SharePoint and OneDrive permissions, so overshared content can be surfaced to anyone who already has access. The risk is that Copilot makes existing sprawl easy to find.

Microsoft's documentation describes how labels and DLP fit in:

  • Copilot honors Purview sensitivity labels and encryption. For encrypted content, users need both VIEW and EXTRACT usage rights before Copilot will summarize or return it.
  • With VIEW but not EXTRACT, Copilot won't summarize the content, but it can return a link so the user can open the item outside Copilot.
  • Some configurations are stricter. Copilot and agents can't access content protected by Double Key Encryption. Copilot also doesn't currently recognize sensitivity labels that protect Teams meetings and chats.
  • A DLP policy using the Copilot location can restrict processing of prompts containing certain sensitive information types, and processing of files and emails with specific sensitivity labels.
  • Microsoft's service description notes that licensing matters. Sensitivity labeling, including automatic or policy-based labeling, requires Microsoft 365 E5 or the Information Protection and Governance add-on. So a tenant without labels may be a licensing decision, not simply neglect. The report doesn't break this down.

So a missing label or DLP policy isn't proof of exposure. It does mean a tenant has fewer guardrails if oversharing exists.

What Microsoft recommends before rollout​

Microsoft's deployment blueprint is built on three pillars: remediate oversharing, set up guardrails, and meet regulations. It recommends that admins:

  1. Identify high-risk sites and files.
  2. Apply interim access restrictions where needed.
  3. Fix access issues and ownership.
  4. Continuously enforce guardrails.

Tools Microsoft points to include SharePoint Advanced Management, Restricted Content Discovery, site access reviews and inactive-site policies, sensitivity labels, and Purview DLP. On visibility, Microsoft says Purview Audit logs Copilot interactions, including which files and sites were accessed. Audit records don't hold prompt or response text, though. For that you need eDiscovery or Data Security Posture Management for AI.

Those last two points address the visibility gap the report highlights, at least for Microsoft's own Copilot. Tools used outside Microsoft's boundary are a separate problem.

A practical checklist for admins and MSPs​

  • Inventory who can reach which SharePoint and OneDrive content. Find overshared or ownerless sites.
  • Check whether labels and DLP exist, are scoped sensibly, and match your licensing.
  • Define one baseline across all tenants, then measure drift against it.
  • Move from annual reviews toward continuous posture checks.
  • Automate policy rollout so a change reaches every tenant in minutes, not weeks.
  • Review Purview audit and DSPM for AI reports regularly.

None of these is proven by the report to reduce incidents. The study measures gaps in practice and doesn't test outcomes.

Reading the commercial angle​

Inforcer's chief strategy officer, Christian Nagele, said businesses will remain exposed until MSPs apply a strong, standardized security posture. He also called Microsoft security and AI one of the biggest short-term opportunities for MSPs, and said those who act within six months could build a lasting advantage. That is a forecast from a vendor whose products serve exactly this market. It isn't a measured result.

The vendor's incentive doesn't make the gaps imaginary, though. Independent MSP coverage this week points the same way. A WindowsForum report on CRN's coverage described similar concerns about Copilot visibility, labels and DLP, and noted that it couldn't locate the underlying report's methodology. That coverage cited a 62% figure for MSPs offering Copilot readiness services, which doesn't appear in the ITPro account. It's a reminder to check the full report before quoting numbers.

Bottom line​

The credible takeaway is narrow: MSPs are selling Copilot quickly, but many lack consistent data protection, AI visibility, and automation across tenants. If you manage Microsoft 365, treat identity and device security as necessary but not sufficient. The data layer, meaning permissions, labels, DLP and audit, is what Copilot touches.

 

References

  1. Inforcer report finds major data governance gap among businesses deploying AI - IT Pro IT Pro 2026-11-19T08:40:00+00:00
  2. Data protection and auditing for Microsoft Copilot | Microsoft Learn learn.microsoft.com
  3. Secure & Governed Data Foundation for Microsoft Copilot - Foundational Deployment Guidance | Microsoft Learn learn.microsoft.com