How does Reality Defender use PII?
Reality Defender uses the media you submit for one purpose: to determine whether it has been manipulated or generated by AI. Our detection models look for signs of manipulation in the media itself. They don't identify the people who appear or speak in it, and we don't use your media to train our models.
Is any uploaded information used to train the AI models?
No. We never use the media you submit, or the results of your scans, as training data for our detection models. We train our models on datasets we license or build ourselves. When customers report a new kind of deepfake, our team studies how the attack works and recreates it with our own data, then uses that data to improve detection.
Do you do voiceprints or ID verification?
No. A voiceprint is a biometric template that identifies a person by their voice. Reality Defender doesn't create or store voiceprints. Our audio models detect whether a voice is synthetic, not whose voice it is.
Reality Defender also doesn't verify identity. Many customers use us alongside an identity verification (IDV) provider: the IDV provider confirms who someone is, and Reality Defender checks whether the selfie, video, or voice they submitted is real or AI-generated.
Is PII filtered from data sent to third-party AI services?
Our detection models run on Reality Defender's own infrastructure, and your media isn't sent to third-party AI services for detection.
The exception is Context Aware analysis. When it's enabled, relevant media data, such as the image or sampled video frames and the filename, is sent to third-party sub-processors such as OpenAI and SerpAPI for contextual analysis. This data isn't filtered or redacted before it's sent. For details, see our current list of sub-processors.
Context Aware is enabled by default. If your organization can't send media to third parties, an admin can turn it off in Settings → RealScan → Enable context models.
What information do you retain, and for how long?
Reality Defender stores the result of each scan, along with basic file information such as file size, length, and GPS coordinates if the file contains them. By default, the media you submit is kept too. Your organization can have it deleted as soon as analysis finishes.
For what's deleted and kept in each product, and how to turn on deletion, see Data retention and deletion after analysis.
How do I delete my account and its data?
See Managing Your Plan & Billing. Once your request is verified, your account and associated data are deleted within 30 days.