How Technology Can Reduce Human Error in Redaction

Tim Leehealey

Feb 27 2019

Legal professionals around the world felt the pain of defense lawyers for Paul Manafort recently, when news media outlets reported the embarrassing news that defense attorneys failed to properly redact draft text from the documents submitted in court filings. Among other things, the redaction error suggested that prosecutors believe that Manafort shared internal Trump polling data with the suspected spy Konstantin Kilimnik during the campaign.

And while the Manafort filing error was caused by attorneys using “masking” in the document, rather than proper redaction steps, there have been other cases in the news recently featuring different kinds of human errors in redaction. For example, last year a court error in Nevada led to the accidental disclosure of the name of a man identified as a person of interest in the deadliest mass shooting in modern U.S. history, the 2017 tragedy at the Mandalay Bay Hotel in Las Vegas.

Other prominent redaction errors that made the news include cases involving Facebook, Pepsi and even media organizations themselves, including The New York Times. “Redaction errors exposing confidential information are coming to light with increasing frequency,” according to Corporate Counsel magazine.

For litigation team members, court staff and corporate employees who tremble at the thought of joining the dubious club of people who have been responsible for redaction errors, there is good news on the horizon: Auto-Redaction is about to become your new best friend.

Humans obviously have a crucial role to play in document review and redaction, but modern review software tools have now evolved to a point where the human role can be secondary — not primary — and can instead be focused on quality control. This is made possible by the creation of auto-redaction functionality in legal review tools, which apply basic critical reactions to all documents in the review set and thereby reduce the risk of any critical information being inadvertently exposed when the documents are produced.

As the name suggests, auto-redaction is the ability to automatically redact a word or phrase across a group of documents — or the entire document — without actually being required to view every single document that is being redacted. This capability is useful in a large number of e-discovery and document filing situations, but it really shines when it is necessary to obfuscate simple words or phrases that are known to be sensitive (e.g., Social Security Numbers, bank codes, names of individuals, etc.).

At the heart of auto-redaction technology is search. If the tool can’t accurately locate the relevant hits, then there is no way it can perform the task. So when evaluating legal review solutions for their auto-redaction capabilities, make sure there are no limitations on its search capabilities. The ability to use wild cards, stemming and other search criteria are all essential.

Filtering is also important. You want to have the ability to target a specific group of documents for auto-redaction, as opposed to redaction against the entire case. This allows for much more efficient and flexible application of the redactions; you don’t want to waste time redacting documents that aren’t at all relevant to the case and won’t ever be produced under any scenario.

You should also scrutinize the redaction options offered by your legal review tool. A good solution should offer the ability to redact multiple words and phrases in a single operation. Beyond that basic capability, it is useful to be able to alter the type of redaction. For example, in some cases, it may be desirable to simply utilize the traditional black redaction box everyone has seen before, but in other cases it may be useful to be able to overwrite the redacted text with alternative text (e.g., overwrite all instances of an individual’s name with the text “REDACTED”). Having this capability in a legal review tool not only allows litigation team members to provide detail as to the reason for the redaction, but it also allows you to use the auto-redaction feature to practice anonymization. This is a critical requirement in many European countries in which individual names are often protected by data privacy laws.

Finally, it is crucial to know what specific Quality Assurance (QA) features the legal review tool provides to your team. If the solution is reliable, most of the time you can trust that the redactions were applied properly, but you still must be able to quickly and easily verify the results yourself. To support this QA, you will want a solution with two key capabilities: (1) The ability to view all documents that have received redactions (i.e., a filter, a label or any other simple method for reviewing a group of documents); and (2) A redaction report feature that shows you where all of the hits associated with the redaction job were found and the context in which those identifications were made by the software.

AccessData’s Quin-C, a next-generation digital investigations and legal review solution, features a robust auto-redaction tool built into the latest release of our product that makes it possible for users to obscure or replace any number of words contained within a document or across an entire case file. This automation capability will be a tremendous time-saver for legal teams who are often required to redact sensitive information from produced documents or court filings, reducing the risk of human error.

Just five years ago, auto-redaction was more of an idea than a widespread technology offering — but today, that is no longer the case. Auto-redaction should become a standard tool in every legal reviewer’s toolkit. It will save you time and possibly save you the embarrassment of inadvertent disclosures that make the news.

Contact us today to learn more about our products and our
approach to improving how you collect, analyze and use data.
Tell Me More