LawSites first reported the figures from the International Legal Technology Association’s newly released survey, which covers more than 500 ILTA-member firms across 12 countries and roughly 140,000 lawyers. ILTA’s public survey page confirms that the 2026 report is its annual firm-technology benchmark, but the detailed results remain behind its member and purchaser access controls. That leaves LawSites as the sole published account of several of the product-by-product percentages, including the 76% Microsoft 365 Copilot figure.
The most defensible conclusion is narrower than the headline: Microsoft’s workplace AI stack has become the common starting point for law firms’ AI programs. It does not establish that Copilot is the most-used tool by individual lawyers, the most valuable tool for legal work, or the product that firms have rolled out most broadly inside their document-management and legal-research systems.
“Using or Exploring” Is a Wide Measure
ILTA’s 94% figure for firms using or exploring generative AI, up from 80% a year earlier, shows how quickly AI has become a mainstream technology-planning issue in legal services. Yet the survey’s wording joins two materially different states: a firm running a paid, tenant-wide Copilot deployment and one evaluating it in a limited pilot both appear in the same bucket.
That distinction changes the reading of the vendor rankings. Microsoft 365 Copilot is attached to productivity software already embedded in most firms’ daily operations: Outlook, Word, Teams, Excel, OneDrive and SharePoint. A firm can explore it as an extension of an existing Microsoft relationship while separately evaluating CoCounsel for research, Harvey for legal work product, Claude for general-purpose analysis, and iManage or NetDocuments tools for matter-document workflows.
LawSites’ firm-size breakdown reinforces this. Microsoft 365 Copilot reportedly led among firms under 50 lawyers, at 59%, but the gap was smaller: Claude reached 42% and CoCounsel 40%. Among firms with at least 700 lawyers, Microsoft 365 Copilot reached 88% while Harvey reached 84%, with Claude at 64%. Large firms are not standardizing on one AI supplier; they are testing multiple systems against different repositories, work types and risk tolerances.
For a CIO, “we are using Copilot” is therefore an insufficient inventory answer. It should be split into at least four questions: whether the firm has licenses, whether users can access the product, whether it is connected to organizational content, and whether attorneys use it regularly for approved tasks. Those are different measurements, and only the latter two begin to describe operational adoption.
The Survey’s Copilot Labels Obscure the Deployment Result
The survey’s most striking deployment claim is that only two products were fully deployed to at least half of a respondent firm’s lawyers: “Microsoft Copilot (formerly Bing Chat)” at 52% and Westlaw Advantage at 50%, according to LawSites. The trouble is that this label is not a clean description of a current Microsoft commercial product.
Microsoft’s current documentation says Microsoft 365 Copilot has been renamed Microsoft Copilot, while Microsoft 365 Copilot Chat has become Microsoft Copilot Chat. Older names continue to appear in some experiences, licenses and capabilities during the transition. Microsoft also differentiates between Copilot Chat, which is primarily grounded in the public web unless a user provides or selects organizational material, and Microsoft Copilot’s paid productivity capabilities that work across Microsoft 365 data subject to user permissions.
“Formerly Bing Chat” points to yet another historical naming path. Bing Chat was Microsoft’s public web AI experience before the Copilot branding era. “Microsoft Copilot for Windows,” meanwhile, describes a Windows entry point rather than necessarily a separate enterprise license or data-grounded service. Treating these labels as interchangeable risks comparing a browser-accessible AI chat experience with a licensed Microsoft 365 productivity service that can retrieve information from a firm’s own tenant.
This is more than a branding complaint. If the 52% deployment figure mainly captures broad availability of Copilot Chat to attorneys signed in with Entra accounts, it says something useful about baseline AI access—but far less about adoption of the paid Microsoft Copilot experience in Word, Outlook, Teams and other Microsoft 365 apps. If it captures a mixture of both, the number cannot reliably serve as a licensing or rollout benchmark.
ILTA has not publicly explained the product taxonomy or published the underlying question wording on its survey landing page. Until it does, firms should not use the 52% number to conclude that a majority of peers have completed a Microsoft 365 Copilot deployment. They can conclude that Microsoft-branded Copilot access is widespread, which is a different proposition.
Why Microsoft Has an Advantage Before a Legal AI Pilot Starts
The survey result does help explain why specialist legal AI vendors’ high visibility at ILTACON did not translate into a top position in the broad “used or explored” ranking. CoCounsel, Harvey and Legora must win a product evaluation. Microsoft often begins inside an estate that firms have already licensed, staffed, governed and connected to identity services.
That advantage is especially pronounced in legal environments where the immediate use cases are document drafts, meeting notes, summaries, research and analysis of long files—the four categories LawSites says appeared at the top of ILTA’s use-case results. Much of that work begins in Word, Outlook, Teams or an existing document repository before it reaches a dedicated legal-research platform.
Microsoft’s enterprise documentation also makes clear why the platform is attractive to firms that cannot treat confidentiality as an afterthought. Microsoft says Copilot honors the Microsoft 365 permissions model, meaning it presents only material the requesting user is already authorized to access. Its documentation says prompts and responses are covered by enterprise data protections for commercial customers, with auditing and eDiscovery support available under the applicable configuration and subscription terms.
Those controls do not cure an over-permissioned tenant. Copilot respecting existing access controls also means it can surface sensitive work product to people who already have access they should not have retained. ILTA’s own 2026 legal-technology programming has emphasized that law firms need classification, access control and data governance before expanding Copilot access. That is the operational issue hidden underneath the vendor ranking: AI makes stale permissions, uncontrolled SharePoint sites and sprawling Teams workspaces easier to discover and use.
Security Has Moved From a Policy Footnote to a Top Concern
LawSites reported that generative AI entered ILTA’s list of firms’ three largest security challenges for the first time and ranked second, behind user behavior but ahead of malware and compliance. No independent public release of the detailed ranking was available, so that placement should be treated as LawSites’ account of ILTA’s executive summary rather than a fully verifiable public statistic.
The finding still fits the implementation reality. A law firm’s risk is not confined to a lawyer typing confidential text into an unapproved public chatbot. Risks also arise from AI retrieving inadequately governed internal content, external agents with their own data-handling terms, copied responses that are never checked, and users mistaking a general web-grounded chat tool for a system authorized to access matter files.
For firms adopting Microsoft Copilot, the priority is to establish a deployment boundary before measuring adoption. That means reviewing SharePoint, OneDrive, Teams and Exchange permissions; applying sensitivity labels and retention policies where they are warranted; defining approved versus prohibited data inputs; enabling logging and review capabilities; and training attorneys on verification, privilege and client-confidentiality duties. A broad rollout without that groundwork may produce impressive access numbers while adding little defensible legal value.
What IT Leaders Should Benchmark Instead
The useful lesson from the ILTA survey is that law firms are moving toward a multi-product AI environment, even when Microsoft is the common foundation. The next internal benchmark should not ask which vendor is “winning.” It should identify where lawyers are actually getting time savings, whether the output is accurate enough for supervised legal work, and whether the firm can explain the data path and controls for every approved tool.
Microsoft’s lead in the survey is evidence of reach, not proof that the Copilot rollout is finished. The firms that get beyond experimentation will be the ones able to distinguish a Copilot Chat entitlement from a Microsoft Copilot license, a pilot from a deployment, and broad availability from safe, routine attorney use.