new image bot/spider/scraper

[lucia]

Lucia, I have a question about method I that you describe.

If Getty comes to my site through picscout, and copies my copyrighted images onto their server so they can then scan them to determine if they are theirs, is that act of copying my images a copyright infringement by picscout?  Or do they have to publish something first?  It seems like they have taken something that is not theirs and made a copy of it.

First that's a legal question and my answer to the legal question is "I don't know". Oscar might know whether their copying onto their server violates copyright or failing that, whether someone alleging copyright would have a colorable case.  I assume picscout would claim fair use for the copy. They might win. Oscar would be the one who could speculate intelligently about that.
Second: They may  not load the images onto their server. 
Third: This might be largely hypothetical since you don't know whether a visit results in their making a copy to their server. They may merely compare. 

If it is infringement, one would have to find a way to prove that this is happening.

Supoena? Discovery after they sue you? Those are the only legal methods I can think of. The illegal ones would be "cracking into their computer".  I'm not sure how you would bring that evidence forward.

I don't want to use the words S.G. hates, but could a poorly designed picscout open Getty up to some sort of large scale action?  They are certainly not teaching or commenting on my photos.

Once again: Legal question.  Worse: complicated fair use question about a transformative use.  (Google gets to copy and cache.  I think that's been deemed fair use.)

If someone steals a priceless painting and doesn't display it, they are still guilty of theft if caught.  Is that also true with digital images?

Well.... Getty seems to say it applies to digital images.

[stinger]

Thanks for those thoughts, Lucia.

[Oscar Michelen]

To my knowledge, Getty does not download every picture it finds onto their servers.  PicScout does the search without having to do that. Looking at images to determine if they are owned or licensed by Getty would not constitute infringement as it is not really a "use" never mind whether its a fair use.   

[Engel Nyst]

I know it's an old topic, just a note about the technical matter here.

PicScout can't look at images without copying them on their servers. It has to download the images.

For example, the technical documentation of TinEye, a similar software, says:
https://services.tineye.com/developers/matchengine/methods/compare.html

If you are comparing by image or filepath then operations are performed in the order in which they are received. If you are comparing by URL then the images are downloaded before the operations go into the queue.

The "operations" may also involve (additional) copying into RAM, but we can ignore that. In order to operate however, all these bots/crawlers first download the images.

I don't question it's fair use, though; just about the technical aspect of it.

[Matthew Chan]

We rarely get into the technical side or detail of how Tineye or Picscout works mostly because it hasn't come up before nor has the discussion been at the forefront.  As such, I don't think most of us at ELI have done research on the matter.  However, it does make for an interesting discussion of how it all works in detail.

But as always, Engel Nyst provides good feedback on the matter.

IN particular, I found this page particularly helpful in describing how Tineye works which I would extrapolate to how Picscout works.

https://services.tineye.com/developers/matchengine/what.html