Reality Defender, a startup developing AI-generated content detection tools, has raised $15 million in Series A funding to expand its team and improve its AI models for identifying synthetic media.
The round was led by DCVC, with participation from Comcast Ventures, Ex/ante, and Parameter Ventures. This brings Reality Defender’s total funding to over $17 million since the company pivoted from nonprofit status last year.
Demand for Reality Defender’s services has rapidly increased alongside the proliferation of AI generative models like DALL-E 2, Stable Diffusion and ElevenLabs. These tools make it easy for anyone to create AI generated images, videos, audio and text.
Reality Defender was founded by former Goldman Sachs VP Ben Colman, Ali Shahriyari, and Gaurav Bharaj. The platform provides its clients, ranging from governments to major financial institutions, with cutting-edge tools designed to detect and stop AI-created content.
The startup uses proprietary deep learning models trained on diverse real-world data to detect manipulated media. It offers an API, web app, and custom solutions to help governments, financial institutions and social media combat deepfake disinformation campaigns.
The company claims its ensemble approach achieve higher accuracy than competitors. But deepfake detection remains an arms race as new generative models emerge. Critics also warn of bias risks in detection algorithms. Reality Defender says its models represent diverse data to mitigate bias.
Reality Defender plans to strengthen its platform with new transparency tools. Later this year it will launch “explainable AI” features that visually highlight manipulated sections of media files, starting with text. This provides intuitive feedback to help users understand the detection process.
The company also revealed an upcoming real-time voice deepfake detection service designed for call centers and fraud prevention teams.
While Reality Defender began as a nonprofit, the company says its transformation into a commercial entity was driven by the scale of the deepfake challenge and the rising commercial demand for deepfake detection technologies.
However, Colman and his team see their mission as not merely business-driven. He expressed their collective belief that, in the face of the perils presented by Generative AI, it is their "civic duty" to counter the worst-case scenarios.
Overall, the new funding enables Reality Defender to stay ahead of developments in generative AI. As faked content spreads rapidly online, AI-generated content detection is critical. Reality Defender aims to be an authoritative solution as more entities take steps to combat potentially dangerous disinformation.
However, Reality Defender faces major challenges in proving its tools can truly keep up with the rapid evolution of generative AI. Recent research indicates many current AI detection systems are easily fooled if text is slightly edited or paraphrased. If the company can meet the formidable challenges ahead and validate its deepfake detection claims through rigorous real-world testing, it will cement itself as a leading bulwark standing between truth and increasingly sophisticated AI disinformation.