[matteason]

How much have you investigated the legality of this? In the EU biometric data is 'special category data' under the GDPR [0] and can only be processed in very limited circumstances unless you have the consent of the data subject [1]

[0] https://www.dataprotection.ie/en/organisations/know-your-obl...

[1] https://gdpr-info.eu/art-9-gdpr/

[vignesh_warar]

Thanks for the info. We don't use any private data, only publicly available images. So it won't be a problem, in my opinion.

I will contact my lawyer and double check this.

[1659447091]

Not only the EU, but you will have to check with each of the 50 US states as they all have a patchwork of laws. Illinois was one of the first, but I don't know much about it; I thought I read it was pretty extensive to the point some facial recognition companies specifically exclude it. Texas also has its own version as well, that I know of; again don't know details.

[matteason]

You might want to look up Clearview AI, who also took publicly available images, performed biometric recognition on them and ended up with a €30.5 million fine: https://blog.barracuda.com/2024/10/23/clearview-ai-fine-gdpr...

[vignesh_warar]

I just did research on this.

Clearview vs Introthem:

- Clearview does photo-to-photo matching. We don't do that, and I don't think I will ever build that.

- You have to provide the name, then we build the faces collection for analyzing at search time and delete it.

- We don't retain any face collection once the search is done.

I still don't know if I am breaking any laws, but here is how Introthem works.

[mathgeek]

You should definitely find out if you are breaking any laws.

[josefritzishere]

100% this is illegal in Illinois. They have specific biometric data protections. Probably also a crime elsewhere.

[knxnts]

yes. and there will be more state laws in the future. seems like it would raise some FCRA concerns to me.

[wongarsu]

The GDPR works on the personally-identifiable vs anonymous distinction. Private vs public doesn't really factor into it, or at least only becomes relevant in the nuances.

Personally identifiable data is just a mouthful, which is why people like to misleadingly shorten it to private data.