Microsoft Word Copilot adds new revision tracking, AI document editing finally leaves traceable and auditable records

ME News Report, April 15 (UTC+8), according to Beating Monitoring, Microsoft CEO Satya Nadella announced today that Word Copilot has received a set of new features aimed at highly sensitive document scenarios. The core change is that Copilot’s editing operations can now enable Track Changes, with every modification recorded with word-level precision, making it reviewable and reversible. Previously, Copilot’s edits in Word were directly overwritten, and users could not audit what the AI changed step-by-step, which was almost unusable in contract review, compliance document finalization, and similar scenarios.

In addition to Track Changes, Copilot also introduces the following capabilities:

1. Commenting feature: can read, reply to, and manage comment threads in documents, with comments anchored to the corresponding text
2. Automatic generation and updating of the table of contents based on Word’s built-in heading styles
3. Management of dynamic page elements such as headers and footers, columns, margins, page numbers, and dates
4. Real-time progress indicators during multi-step editing, informing users which step is currently being processed

These features are based on Microsoft’s Work IQ layer, complying with Microsoft 365’s data security boundaries, retaining sensitivity labels, and enforcing data leakage prevention policies. Currently, they are only available on Windows desktop through the Frontier program, with Mac and web versions to follow later.

Track Changes is the most critical update in this release. Legal, financial, and compliance teams often require complete audit records of modifications in documents, which are mandatory for regulation and internal processes. Copilot previously lacked this capability, meaning these teams either had to avoid using AI or manually compare changes paragraph by paragraph after use. Now, Copilot works directly in review mode, with editing traces formatted identically to human edits, allowing seamless integration into existing review workflows.

(Source: BlockBeats)