Prevent Content Leaks with Digital Watermarking

How It Works

Introduction

Every company and government wants to keep its digital assets and data private and secure. However, time and again, we see news headlining recent leaks. For companies, when these leaks occur, they lose control over their own marketing campaign and roadmap secrecy, the morale of employees is negatively impacted, and the costs to the company can be in the millions of dollars (for example, the recent Rockstar GTA 6 Leak cost Rockstar $5 million and thousands of hours of staff time). For governments, the impact of leaked digital media and data can be even more costly (Pentagon document leak containing details on Ukraine War could kill (usatoday.com)).

To prevent data leaks, companies and governments spend an enormous amount of money each year on digital security and data loss prevention. Yet, the leaks continue to happen. Why? There are many reasons, but one of the primary missing pieces in current data loss prevention systems is the ability to trace the leaks when they happen. This makes sense, right? How can you plug a leak without knowing the source? If your car tire is leaking air, you put soapy water on it to see where the bubbles come from. Without knowing the source, you can’t possibly hope to plug the leak.

This is where Steg AI comes in. Steg AI offers a state-of-the-art solution to leak tracing and prevention. In particular, we use our in-house deep learning technology to invisibly watermark digital assets with unique identifying information that allows us to trace the source of leaks if and when they occur; and, we integrate with Drives (e.g., OneDrive, Google Drive) and Digital Asset Management (DAM) companies to keep the process simple and to keep the digital assets in their original location.

Steg AI’s Invisible Watermarks

Steg AI uses proprietary deep learning technology that we develop in-house to invisibly watermark digital content (e.g., images, videos, documents, audio, etc.). These watermarks perturb the data itself to bake information into that data. That information can be anything the user wants; for example, basic information like author and license and complex information like the meeting in which the asset was shared and the people that were there.

The figure below shows an example of the watermarking technology being applied to an image. The image on the left is the original and the image on the right is the watermarked version.

Steg AI Watermarking Example: The left image is the original image. The right image has been watermarked using Steg AI’s proprietary watermarking technology.

Steg AI Watermarking Example: The left image is the original image. The right image has been watermarked using Steg AI’s proprietary watermarking technology.

With images, the pixels represent the underlying data and our watermark is baked into those pixels through small perturbations in intensity values. One big positive of this approach is that the watermark is nearly impossible to remove. For example, removing the metadata of the image has no impact on the watermark whatsoever, and the watermark will survive more complex image manipulations like cropping, enveloping, occlusions, blur, color shifts, etc.

For more information on Steg AI watermarking, check out the Technology section of the Steg AI website.

Using Steg AI Watermarking Technology for Leak Tracing

Steg AI watermarks enable tracing the source of leaks. The key is that each time digital assets are shared, a unique, traceable, new watermark should be applied to those assets. In this way, each watermark adds a version to the asset that can later be traced back to where that asset was first shared.

The figure below outlines a flow we find often works for our customers. In particular, most companies use some Drive (e.g., Google Drive, OneDrive, etc.) or Digital Asset Management (DAM) system to securely store their digital assets. However, during the lifecycle of a digital asset, it needs to be shared with many internal people and teams and possibly even with external teams. Steg AI will integrate with these Drives and DAMs so that each time an asset is shared, we can watermark that asset, applying a new unique identifier and the required metadata related to the teams involved, the meeting where it was shared, etc. such that we can later trace a leaked version back to the source. In the figure, the assets from the Drive/DAM are watermarked with a unique identifier for the internal “orange” team, a different unique identifier for the internal “red” meeting, a different unique identifier for the external “blue” company, and so on.

Now, each of these different people/teams/companies have their own version of the asset. They all look the same visually (because the watermark is invisible), but the watermark is different for each recipient.

Watermarking Example. Steg AI integrates with your Drive or DAM (or you can use Steg AI’s own dashboard). Then, every time you share an asset, Steg AI will apply an invisible, unique, traceable watermark to those assets. In the example above, different watermarks are applied to the internal orange team, to the assets that are shared in the internal red team meeting, and to the external blue company.

Watermarking Example. Steg AI integrates with your Drive or DAM (or you can use Steg AI’s own dashboard). Then, every time you share an asset, Steg AI will apply an invisible, unique, traceable watermark to those assets. In the example above, different watermarks are applied to the internal orange team, to the assets that are shared in the internal red team meeting, and to the external blue company.

In the event of a leak, we can then download the leaked digital assets, read the watermark and metadata, and trace the leak to the source. In the figure depicted by the figure below, a new product was leaked on a news outlet called “The Leaky Pipe”. Steg AI then downloads the image of the product and reads the watermark, leading us to the source: the red team meeting.

Leak Tracing Example. In this example, a photo was leaked of the thingamajig machine 4. Steg AI downloaded the image and read the watermark. The watermark points back to the red team meeting as the source of the leak.

Leak Tracing Example. In this example, a photo was leaked of the thingamajig machine 4. Steg AI downloaded the image and read the watermark. The watermark points back to the red team meeting as the source of the leak.

Next Steps

There are many ways that the above solution can be adapted to different customer scenarios. Steg AI watermarking and leak tracing products can be surfaced through our API or Web Application or via integrations with Drives/DAMs or common software workflows like Photoshop plugins. If you are interested in learning more about leak tracing and prevention using Steg AI technology. Please contact us.