Thank you for joining us again for the fifth chapter of our AI Redaction series. Now that we’ve covered the basics of redaction, redacting documents with AI, the best practices for redaction and how to implement them, and how automated redaction software and services work, we turn our attention to the ins and outs of video redaction.
Summary
- What challenges does video redaction present?
- Which industries need to redact confidential information from video footage?
- How can AI-powered redaction software help solve the challenges of video redaction?
Video editing for any purpose is a time-intensive process–even a 30-second social media video could take hours to put together. Redacting confidential information and personally identifiable information (PII) from video evidence and video surveillance takes even more time. This is especially true when teams need to review entire video files, identify sensitive data, and redact parts of a whole video without compromising accuracy.
However, in order to keep the PII of innocent bystanders and law enforcement personnel safe, video redaction is crucial. In many cases, teams must also support audio redaction to protect spoken names, addresses, and other personally identifiable information that may appear alongside the visual record. AI-powered video redaction software starts to remove the human element, taking the brunt of the workload in regard to video redaction, and freeing up resources for law enforcement and justice agencies.
The challenges of video redaction
The average video contains 30 individual frames per second of video footage, and when manually redacting PII from video evidence, law enforcement personnel and members of justice agencies have to edit videos frame-by-frame. A 30-second video could contain as many as 900 frames or more, and an experienced technician could take as many as eight hours or more to redact PII in a 10-minute video clip.
This makes manual redaction one of the most time-consuming parts of handling audio and video evidence. Compared with traditional methods, modern video redaction tools can dramatically reduce redaction time, helping agencies save time while maintaining consistency across large volumes of footage.
When a human technician redacts video evidence, there is an increased risk of human error. For example, a technician may miss an innocent bystander’s face in the background, or a license plate, or a badge number? Someone’s privacy, or even safety, could be at risk.
Unfortunately, not all video evidence that needs redaction is pleasant to look at. If a human technician has to redact gory or disturbing video evidence, like footage of a violent crime, repeatedly viewing such traumatic images could have a negative effect on that technician’s mental health and quality of life.
AI-assisted video redaction exists to remedy these issues in a plethora of ways. A typical workflow includes ingesting footage, identifying sensitive content, choosing how to apply redaction, reviewing the results, and exporting the final file so the obscured areas are rendered irreversibly. It can accelerate the video redaction process, saving up to 90% of the time it would take to manually redact video evidence. AI-powered video redaction software can reduce the mental strain on human technicians, catching PII they might have missed or redacting PII from graphic content that might take a toll on an individual’s mental health.
Through automatic detection and object tracking, the right platform can identify and follow faces, people, license plates, and other objects across frames, making it easier to blur faces, blur vehicles or identifiers, and easily redact sensitive footage from CCTV footage, dash cam, or body worn video.
Video redaction in different industries
Video redaction is used frequently by law enforcement agencies, in order to prepare video evidence for viewing by different populations—for use by news agencies or as evidence in the courtroom. It is also increasingly important for government agencies responding to public records requests and FOIA requests, where footage may need to be reviewed, redacted, and shared while still protecting sensitive information. In either case, it is imperative that this video evidence be devoid of any PII that isn’t related to the focus of the evidence. Bystanders’ faces and license plates need to be removed, as well as any other information that could be used to identify a witness.
Security agencies and legal entities, who generally prepare and submit video evidence for purposes similar to law enforcement, have similar responsibilities to preserve the safety of PII. Maintaining confidentiality of sensitive information is also of the utmost importance, and redacting video evidence can also help avoid legal liability.
Legal teams may also have to compile and redact data to provide the correct information to fulfill a Freedom of Information Act (FOIA) request. In these cases, agencies may need to download, review, upload, and deliver redacted public records under strict timelines, making efficient workflows and reliable access controls especially important. Data is redacted in these types of requests in order to protect personal privacy, maintain confidentiality, ensure compliance with the law, and preserve the integrity of the legal process. As we detail in Chapter 2 of our blog, legal documents fall under a specific set of redaction constraints.
In video contexts, these categories may appear visually, verbally, or both, which is why many organizations need tools that can handle both footage and related audio evidence.
Types of information commonly redacted:
Personally Identifiable Information (PII)
- Names, addresses, phone numbers, email addresses, social security or driver’s license numbers, medical or financial information, and more
Confidential or proprietary information
- Financial information, domain names, trade secrets, copyrights, ideas, techniques, inventions, and more
Financial information
- Financial transactions or statements, bank account or credit card numbers, credit ratings, payment histories, and more
Medical or personal health information
- Medical history, treatments, conditions, insurance information, and more
Information related to ongoing legal proceedings
- Ongoing litigation or settlement negotiations, information that could be prejudicial to the outcome of a legal case
Sensitive personal information
- Information about an individual’s sexual orientation, religious beliefs, or political disposition
Children’s information
- Information, such as names and addresses, of minors
Video redaction software and tools
Video redaction software can speed up the redaction process by a significant margin. The best video redaction software combines intuitive tools with automation, allowing users to review footage, select the right redaction method, and process files more efficiently than with manual workflows alone. Veritone Redact automates the video redaction process by automatically tracking sensitive imagery and detecting relevant items such as heads, license plates, officer notepads, and laptops. Depending on the use case, teams can obscure content through blurring, pixelation, blacking out, or solid masking, helping ensure that sensitive areas cannot be uncovered later. This functionality all but eliminates the need for a human technician during this tedious portion of the redaction process, meaning nobody has to take on the mental burden of obscuring the faces in a video of a violent crime or other traumatic incidents.
If an individual user needs to redact data not automatically detected by the aiWARE platform at the core of Veriton Redact, that user has the option to manually define objects to obscure. Veritone Redact will then allow the user to use their cursor to follow that item throughout the video or opt to let the AI-powered redaction software track the defined field through the rest of the video.
This flexibility is important when users need to redact highly specific details, including specific words, documents visible on screen, or other edge cases that may require review outside automatic mode.
Unlike many AI-powered video editing tools available, Veritone Redact is developed specifically for use by public safety and justice agencies, meaning it comes equipped with functionality to aid in chain of custody requirements, collaboration between users, and maintaining audit logs of all actions taken.
In addition, Veritone Redact can come as part of Redaction Managed Service, a white glove managed service aimed at providing professional support for law enforcement agencies and legal organizations without requiring them to invest in new technology. This service handles maintaining, monitoring, and reporting via its AI-powered redaction software, allowing quick delivery of quality, professionally redacted files to clients. Redaction Managed Service also helps with FOIA compliance by quickly making redacted records available, helping build community trust in local law enforcement and legal agencies. This is particularly valuable for organizations that need bulk redaction support or lack the staff to manage large volumes of incoming footage efficiently.
Conclusion
Thank you for continuing to join us for our AI redaction series. Here, we discussed the challenges of video redaction, focussing on how time-consuming the process is and the toll it has the potential to take on the technicians manually obscuring the video. AI-assisted video redaction software can cut the time to redact a video by 90%, which comes out to massive time savings when it can take up to eight hours to redact a single 10-minute video.
We detailed the different industries that might make use of video redaction software, from security to the legal and law enforcement fields, and the different benefits video redaction software offers to each type of organization. In addition, if you need a reminder of the type of information commonly redacted from files, it has been summarized.
Lastly, we outlined the capabilities of AI-powered video redaction software like Veritone Redact and Redaction Managed Service, a white-glove managed AI-powered redaction service. For teams handling audio and video evidence together, this kind of workflow can also help bridge the gap between video review and the need to support audio redaction in related investigations or disclosure requests. In our next chapter, we’ll discuss redacting audio files. If you or your team are interested in AI-powered automated redaction software or redaction services, contact a Veritone team member to learn more.
