[[CLIENT_1]] will acquire the surgical robotics division of [[COMPANY_2]] for $47.3 million, with closing scheduled for September 14, 2026, following regulatory approval in Germany.
The client’s name is gone. But is the client protected?
Someone familiar with the industry may recognize the transaction from its exact value, business division, closing date, and jurisdiction. Each detail appears harmless by itself. Together, they may point to one specific matter.
Removing direct identifiers is not the same as reducing identifiability.The false confidence
A protected document can still reveal the matter.
Names, emails, addresses, and account numbers are the easiest sensitive details to recognize. Replacing them is an important first step, but legal documents also contain facts that become identifying when they are combined with industry or matter knowledge.
Northstar Medical Ltd. will acquire the surgical robotics division of Calder Group for $47.3 million, closing on September 14, 2026.
[[CLIENT_1]] will acquire the surgical robotics division of [[COMPANY_2]] for $47.3 million, closing on September 14, 2026.
[[CLIENT_1]] will acquire a business division of [[COMPANY_2]] for [[AMOUNT_1]], closing on [[DATE_1]].
All organizations and facts in this example are fictional. The reviewed version is more limited, not automatically anonymous or suitable for every AI use.
Three layers
Review what is explicit, contextual, and hidden.
Direct identifiers
Names, contact details, addresses, account numbers, registration numbers, and signatures.
Often detectable from the text itself.Contextual identifiers
Exact values, dates, locations, products, project names, unusual obligations, and distinctive facts.
Often sensitive only in the context of the matter.Hidden identifiers
Metadata, comments, track changes, revision history, filenames, file paths, and document properties.
May not appear in the visible document body.The identifiability test
Ask four questions before the prompt.
- 01
What could a stranger identify?
Look for names, contact details, addresses, and standard identifiers.
- 02
What could someone in this industry identify?
Consider transaction values, products, dates, locations, regulatory events, and publicly reported facts.
- 03
What could someone inside the firm identify?
Review matter names, client aliases, codenames, strategies, and distinctive internal terminology.
- 04
Which remaining details does the AI need?
If a precise date, value, location, or name is unnecessary for the task, replace or generalize it.
What is the minimum information required to obtain a useful result?
The human boundary
Software can suggest. Matter knowledge must decide.
Automatic detection can find common identifiers and suggest people, companies, dates, addresses, and financial values. It cannot know every fact that is sensitive inside a firm.
It may not know that an ordinary phrase is a confidential project name.
It may not know that a product description identifies the client.
It may not know that an exact amount is recognizable within a small industry.
It may not know that a fact pattern has already appeared in the news.
Where LegisComply fits
Make review a visible step, not a promise.
LegisComply processes the document locally, detects likely sensitive details, and lets the user confirm, ignore, recategorize, or add every replacement. The original values and the mapping used to restore them remain in the browser tab.
Detect common patterns and contextual entities locally.
Add matter-specific terms only your team would recognize.
Review the protected working copy before it leaves the browser.
Use the result only with an approved AI and within applicable policy.
Try the local review workflowThis guide provides general educational information and does not constitute legal advice.