San Francisco, April 13, 2026, 05:10 PDT
Anthropic’s Claude is now available in beta for Microsoft Word, rolling out over the weekend as a sidebar tool. Users get AI-powered answers about their docs, can have text rewritten, and see all edits tracked by Word’s built-in system for easy approval or rejection. Legal contract review tops the example use cases on Anthropic’s help page.
This release takes Anthropic further into the highly competitive enterprise AI space, specifically legal services. Law firms, feeling the heat to keep up, are changing their tune. McCarthy Tetrault partner Susan Wortzman told Reuters last month that where firms previously assured clients they weren’t touching generative AI, “now they are saying you must use it.” Reuters
That puts Anthropic squarely in Microsoft’s main software domain. Back in late March, Microsoft announced its Researcher agent in Microsoft 365 Copilot would support Claude models, with executive Nicole Herskowitz telling Reuters that customers are asking for “various different models from different vendors” within Copilot. Now, Anthropic is rolling out its own integration for Word, adding another option to the mix alongside its existing Microsoft partnership. Microsoft Support
Anthropic’s new add-in rolls out to Team and Enterprise users on Word across the web, Windows, and Mac. Team seats are listed at $25 per user each month, billed monthly, according to Anthropic’s pricing page. While working through comment threads or drafting in templates, users can flip between Sonnet 4.6 and Opus 4.6, with numbering and styles staying intact.
Anthropic isn’t pulling any punches with its legal use cases. Sample Claude prompts run the gamut: summarize governing law, spot commercial terms that stray from the norm, generate indemnity language, or carve out liability clauses—mutual or otherwise. It’ll also break down which redlines on edits are flagged as dealbreakers, and which terms are flagged for review.
Claude for Word now faces off with legal AI specialists like Harvey and Legora, as well as established names LexisNexis and Thomson Reuters, who fold AI into their own case-law and workflow packages. Last week, Reuters reported Harvey’s valuation hit $11 billion following its latest funding round, while Legora reached $5.55 billion after its own raise. Both are pitching their tools to law students, betting on future demand as the legal AI sector pushes toward $10 billion in annual value by 2030, according to some estimates.
Speed doesn’t replace the need for review. AI “hallucinations”—cases where these systems make up facts or sources—already have prompted courts in the U.S. to scrutinize or sanction attorneys at least seven times, according to a February report from Reuters. Suffolk University Law School dean Andrew Perlman didn’t mince words either, calling it “incompetence, just pure and simple” to use generative AI for citations without verifying them. Reuters
Anthropic has run into this issue itself. Last year, Reuters reported that a Latham & Watkins attorney representing the company in a copyright lawsuit told a federal judge that Claude, the company’s AI, generated an incorrect article headline and author names for a genuine source. The judge called it a “very serious and grave issue.” Reuters
Anthropic’s own docs highlight another wrinkle: prompt injection—basically, hidden cues stashed in text, comments, or tracked changes that might nudge the model off course. The company specifically cautions users against running the add-in on documents from untrusted sources. It also makes clear that the beta isn’t intended for client-ready work or legal filings unless it’s been checked by a human. There’s no chat history between sessions, and right now, the tool doesn’t connect to Enterprise audit logs or the Compliance API.
Claude for Word isn’t running solo. It taps into shared context with Anthropic’s Excel and PowerPoint add-ins, pointing to a push for Claude to tag along as users move between docs, spreadsheets, and presentations—no more keeping it isolated in a chat box. Back in March, Anthropic said it had already updated its Excel and PowerPoint tools to sync context between applications. The company’s Word help page confirms this: the same cross-app workflow now connects all three.
Law firms aren’t debating if AI will land inside Word—they’re deciding how much trust to put in it once it’s there. Anthropic is pushing the angle that built-in convenience could edge out niche players, yet with legal work demanding verifiable outputs, proprietary content, and strict oversight, specialists aren’t out of the picture just yet.
