Frequently Asked Questions

Reality Defender is a deepfake detection platform for enterprises, governments, and platforms to detect AI-generated content and manipulations across audio, video, image, and text files.

No. Reality Defender is meant for the largest entities and governments to potentially shield hundreds of millions from deepfakes. We believe the onus of deepfake detection should not be on the end-user, consumer, or citizen, but on the largest institutions and platforms serving them.

No. Watermarking requires provenance and ground truth, which requires buy-in from all generative content models. As such a feat is impossible, we use an inference system that grades each content with a 1-99% probability rating -- all without the need for the ground truth.

We are not leveraging Generative AI technologies. Our detection models are a combination of in-house models or proprietary models.

The platform has a range of US, UK, and EU certifications including SOC2 Type 2 and GDPR.

We use state-of-the-art neural networks (CNNs, Transformers, ViTs, including large foundation models) to model discriminating features that help differentiate generative media from real media. In this process, we perform spatial, temporal, and frequency domain analysis along with domain specific feature losses (such as artifacts in images, etc.). Furthermore, we support multiple models for various modalities.

We have created a diverse in-house dataset consisting of videos, image, audio (including telephone quality) and text.

What a contact-center tool has to do: score live audio in real time inside the call flow, catch both synthetic voices and AI-agent callers, and give agents a verdict they can act on

Frame it as call screening at the IVR, before routing. The question that matters is whether the caller is a human or an AI agent

Explain the API pattern: stream or submit the audio, get back an AI-agent verdict, act on it in the routing layer