What if I find a file that is flagged incorrectly?
Attestiv provides several calibration profiles that can be selected depending on the type and source of items being analyzed. Selecting the appropriate setting will often help properly calibrate your system and eliminate false positives or negatives. However, if Attestiv continues to flag files incorrectly, please bring it to our attention via our support web page. […]
Why are a lot of my photos flagged as suspicious?
Often photos are manipulated or resized as part of capture, handling or to try to save storage space. In such cases, the photo may not have been maliciously altered but will still contain traces of manipulation. Attestiv offers tunable settings through the dashboard and can be configured to selectively ignore common manipulations that are not […]
I’m looking for a photo of a screen or screenshot detector. Does Attestiv detect these?
Attestiv detects photos of screens, screen captures, and photos of photos.
I’m looking for a Photoshop detector. Will Attestiv flag Photoshop and other photo editors?
Yes, Attestiv detects various photo editors via traces and pixel-level analysis.
I’m looking for a generative AI detector. Does Attestiv flag generative AI?
Attestiv flags generative AI produced by many of the popular text-to-image platforms, including Dall-E, Stable Diffusion, and Midjourney. As you may have noticed, new versions and new platforms appear regularly, and we strive to be up-to-date on the latest. If you discover a new generative AI framework that is not supported by our analysis, please […]
What is a false positive?
A false positive occurs when an authentic item is incorrectly flagged as suspicious.
What is a false negative?
A false negative occurs when a suspicious item is not flagged or missed by the system.
What is AUC?
AUC stands for the area under the curve, which provides an aggregate measure of accuracy across various classification thresholds. Specifically, it takes into account accuracy relative to false positives and false negatives.
What is the accuracy of the Attestiv platform?
Each model used by the Attestiv platform is tested to a minimum of 97% AUC and biased toward false negatives to try to avoid false positives. However, each model is trained on certain types of anomalies or potential fraud and is not all-encompassing. We strive to identify new threats and include them as part of […]
How many models are used to generate the tamper score?
The tamper score aggregates numerous models that run on each item to produce the score. There are over 80 models Attestiv uses across a number of categories listed on the Attestiv website technology section.