Hello and welcome to the third chapter of our AI Redaction series, which will cover redaction best practices for legal professionals. Chapter 2: Redacting Legal Documents With Artificial Intelligence was all about the types of information that are typically redacted from legal data (such as personally identifiable information, or PII) and the challenges that legal and eDiscovery teams run into during the redaction process.

In this chapter, we’ll get into the best practices for legal data redaction and why it is such an important topic for legal professionals to stay up-to-date on.


Lawyers, eDiscovery teams, and other legal professionals handle mountains of sensitive data on a daily basis. Not only are they handling and managing this information, but they are also responsible for providing information to the courts, regulatory agencies, opposing counsel, and sometimes citizens who have made requests for governmental records or personal data. Learning about and integrating redaction best practices are great ways to ensure that legal teams are sharing all the necessary information without breaching privacy or confidentiality.

What needs to be redacted in legal documents?

When it comes to redaction, legal documents and data have a specific range of information that should remain confidential in order to protect the individuals involved and maintain legal compliance. 

We have detailed information on what pieces of information should be redacted from legal documents and data in Chapters 1 and 2, but as a refresher, private, confidential, or privileged information can include:

  • Personal information, which can include PII like their name, Social Security Number, date of birth, protected health information (PHI), or some financial information.
  • Privileged information, such as information that’s protected by the attorney-client privilege as attorney work product.
  • Confidential business information regarding intellectual property, trade secrets, internal organizational strategies, etc.

Depending on the sensitive information, its disclosure may occur in the context of eDiscovery, court filings, or at other times during the course of litigation. Legal teams may also have to compile and redact data in order to provide information to fulfill a Freedom of Information Act (FOIA) request, or state law corollaries like open records laws, public records laws, or sunshine laws, which enable citizens to enquire and request information about how their federal, state, and local governments operate.

As stated before, the information that is being redacted will depend on whom this data is being shared with, but regardless of the recipient of the information, the data should be removed or redacted thoroughly. This could look like blacking out text, blurring someone’s face, or even removing someone’s name from an audio file—all of which require a lot of time and concentration when being conducted manually.

What goes into the manual redaction process?

The manual redaction process is a time-intensive one. For redacting documents, this could look like taking a black marker or a digital redaction tool and covering up the necessary text. For photos and videos this could look like pixellating or blacking out PII, someone’s face, etc. And for audio and video, this could mean bleeping out or cutting moments where PII or other confidential information is mentioned.

While the process sounds straightforward, it’s incredibly time-consuming when done manually. Aside from needing to check and double-check the work and creating a redaction log to maintain compliance and chain of command, there is a large window for human error and inadvertent disclosure, both of which could be detrimental and seriously harmful to the case, individuals involved in the case, and the legal professionals who are processing the information.

This is especially true when video is involved. When a video needs to be redacted manually, someone must go through the video frame by frame to ensure that nothing is missed—meaning that a 10-minute video could take an entire week to complete properly. Plus, on top of this time constraint, practices and teams need to also consider the cost it takes to dedicate employees’ time to these efforts.

Fortunately, there are redaction software and tools available to help eliminate the need for manual redaction, enabling teams to speed up the redaction process while increasing accuracy.

What are the best practices for automated redaction?

Implementing automated redaction software is by far the best investment a company can make when it comes to cutting costs and time loss due to lengthy redaction processes. We’ve touched on what’s involved with manually redacting specific files, now let’s discuss best practices for automating the redaction of these formats.

Depending on the file’s format, different best practices could apply. For physical documents and images, the redaction process will obviously remain manual, but depending on the tool in use, scanning the files could enable automation opportunities.

Best practices for text document and PDF redaction

The simplest way to redact information from documents and PDFs is to omit the information altogether. With automated redaction, teams can choose or input specific keywords, names, numbers, etc. that need to be redacted and the software can take care of it from there.

Best practices for audio and video redaction

Although the same concept and idea applies to audio and video redaction, due to the nature of these mediums, these files require a more intensive, more specialized process, which can greatly benefit from automation. 

With audio redaction, automation software can identify pre-determined keywords and pieces of personal or confidential information and remove them from the audio. This becomes exceptionally useful when that audio is also part of a video file.

As mentioned before, video presents a huge time commitment: a five-minute video rolling at 30 frames per second requires redacting 9,000 frames. With automation software, legal teams can set their desired redaction needs, and AI can perform this tedious task in a matter of minutes while still maintaining a clear redaction log.

Using Veritone Redact for automated redaction

With Veritone Redact and Redact Managed Service, legal firms can save time and money while operating at a higher speed, efficiency, and accuracy. Thanks to our proprietary aiWARE platform, the reduction process can be performed in a matter of minutes as opposed to days. Plus, users still maintain total visibility and control—with Redact, it’s easy to input which pieces of information need to be redacted, and the review process is just as clear and straightforward. 

Check back for our next chapter for more information on automated redaction with AI.


Thanks again for joining us for our third chapter of this AI redaction series. We’ve discussed what types of information typically need to be redacted from legal data, the pitfalls of manual redaction, the best practices for automated redaction, and how AI-powered tools and services like Veritone Redact and Redact Managed Services can help improve legal teams’ speed and accuracy. Join us for Chapter Four, where we’ll dive deeper into automated redaction using artificial intelligence.

If you or your legal team are interested in AI-powered automated redaction software or redaction services, one of our Veritone team members can help you learn more about our offerings.