The Reality Defender team recently had the opportunity to share our detection methods and industry outlook with Gartner as part of the firm’s Emerging Tech report (paywall). The report addresses the escalating issues of disinformation propagated through AI-generated deepfakes, while also underscoring the need for robust, multi-faceted detection models to combat the growing threat of AI-driven dangers.
The ease of use and availability of generative AI tools — including and especially those used for fraud and disinformation — requires an equally (if not more) robust and scalable slate of detection methods that can analyze a near-infinite pipeline of content in real time. Experts predict that as much as 90% of online content may be synthetically generated by 2026. Leveraging multi-model tools using AI to identify AI has become increasingly crucial as the majority of online spaces become filled with deepfake “slop.”
Gartner's Recommendations
Because malicious actors constantly strive to create increasingly more sophisticated deepfakes to elude detection, the struggle between the forgery and detector has turned into an arms race that requires a multi-layered approach. As per Gartner’s findings, the report recommends a combined utilization of established and emerging technologies to detect deepfakes, including (but not limited to) machine learning and deep networks. Reality Defender’s detection platform employs an ensemble of models to identify fake/manipulated content across multiple modalities, simultaneously using patented techniques and leveraging the aforementioned technologies (among many others) to give the most robust and complete data for every file scanned.
Gartner’s report ultimately recommends companies and clients invest in robust multi-modal detection technologies and regular updates for these systems to address new generative AI applications and to ensure success against emerging threat vectors. The firm’s experts emphasize the importance of clearly explaining why content is classified as tampered or forged to enhance transparency and trust in the technology. Reality Defender’s detection suite provides such clear explanations via a cumulative score, which determines the likelihood that media in question is or is not a deepfake, along with a wealth of additional data that affords our clients to make informed decisions and take action if need be.
Gartner clients and subscribers can read the Emerging Tech report in full here. The Reality Defender team is grateful to our colleagues at Gartner for their crucial work and insights, as well as the Gartner team’s desire to collaborate with emerging research-driven platforms.
