Though much of what’s in store for 2024 remains uncertain, Reality Defender is well-prepared. Since 2021, our team has planned far ahead for anything and everything this year might bring, investing heavily in ensuring our deepfake detection platform can rapidly adapt to new synthetic media techniques and distribution tactics that may emerge.

As the best-in-class deepfake detection, we are already entrusted by enterprises, governments, and institutions to instill trust and clarity in the age of AI-driven disinformation. We continue to further solidify this trust and remain the best with a constant stream of updates, upgrades, improvements, and iterations on our platform and detection models.

Image Detection with Explainable AI 

Our most recent update focuses on sizable upgrades and improvements to image detection. As deepfaked and AI-generated images become infinitely more convincing, our models have responded in kind with upgraded capabilities, including Explainable AI.

Like Text Detection before it, Explainable AI with images provides visual heatmaps to highlight areas in images that contributed most to our models’ result. For instance, if an area of an image is highlighted in red, this highlighting is essentially circling the part of the image that made our models call the image as a likely fake. This helps give greater transparency as to how our Image Detection works, what it looks at, and why it gave a certain result.

Preprocessing and Platform Upgrades 

Also launching in this update is our intelligent preprocessing pipeline, which swiftly identifies key faces for analysis by drawing bounding boxes around faces over 100px  by 100px. Now, each face in every photo and video receives predictive scores from multiple deep learning models. The scores are then blended into one easy-to-understand ensemble score, giving clients robust results with more data.

The Reality Defender web application and API have also added support for analyzing YouTube Shorts — a highly-requested user feature. This addition is small but mighty, allowing us to launch related improvements down the line. 

This year will see advancements in AI that will blindside even those with the greatest foresight. It will surface new challenges that test our vigilance. Major elections in the U.S. and abroad will undoubtedly see an uptick in synthetic media created and distributed with intent to harm, while bad actors and state-sponsored attackers will continue to improve their weaponization of generative AI.

Yet in the face of uncertainty and disinformation, Reality Defender stands firmly on the side of truth. We will continuously enhance our deep learning models to not only keep pace with AI's rapid growth, but model ahead for what it may bring down the line.

The future remains unwritten. Even with rampant uncertainty, we truly believe 2024 can mark a greater turn toward truth.