Quote:
Originally Posted by cerulean
Does Go.cam have a whitepaper like Yoti? You said that there was evidence and test data but I don't see where you've posted it.
When I asked Go.cam months ago about my concerns with face-api.js and tesseract.js, both used in their open source implementation, I was met with a very business-heavy hand wave, which pushed me away from them and toward other companies in this space.
I like that it's continually being developed, but I don't see any concrete information about what they're improving.
Yoti's white paper suggested that there was as high as a possible 10+ year gap in age detection via age estimation. In those cases, my understanding is that it falls back on ID scan + facial recognition.
I am not sure I fully trust certification companies, because when I reached out to them about what their tests were, they never got back to me.
I also saw that Go.cam isn't fully free, which makes sense honestly, because of this line:
Without live presence detection, I don't see how we can be sure that this isn't just a fake webcam or an AI generated face.
My clients speak to their lawyers, and their lawyers can tell them how to prepare for a lawsuit in this area or how to protect themselves as best as they can, but they do not have intimate knowledge of these systems. When I'm asked whether a system like this can do what their lawyers say protects them the most, it's very difficult for me to say yes or no without transparent and plentiful information.
It would be nice if AVS companies would start carrying insurance for lawsuits. Now that would put my clients minds at ease.
|
GO.Cam is certified by Agecheck and KJM results of age estimation accuracy test on a 10k+ open data set is available. It can be found on the github project and for 100% transparency the very simple Phyton code to reproduce the results is also included in the folder. We encourage you to run it.
The advanced liveness feature is currently only available in the hosted version currently. The hosted version is free.
The website operator can choose the level of liveness. Advanced liveness is more secure and a bit longer. In both cases videos and static pictures are detected.
About Insurance.
Go.cam is a free and open-source tool. You are free to use it, adapt it, and even audit it line by line if you wish. This also means that we do not provide it with legal assurance: we deliver the technology, but you remain responsible for how you integrate and use it within your compliance framework.
Our role is to ensure that the technology is functional, transparent, and continuously improved. How it is used, integrated, and maintained within a legal or compliance framework remains the sole responsibility of the website operator. In the github project you can find the on device AI age estimation evaluation method.
Today, hundreds of websites around the world already rely on go.cam with around 150k verifications per day on the hosted API and much more using the self hosted open source to provide secure and reliable age verification.