They don't need to. They can just cheaply crawl the image sites for those informations. Easier. Less resource needed. No community to manage.

Not only that, but they already have cloud environments that allow them to get all of that data anyway. Android phones come pre-configured to automatically upload every photo you've ever taken to the cloud, Amazon offers unlimited personal photo storage, and Google and Amazon offer the two largest enterprise cloud computing environments available, so they can likely just analyze imgur's logs.

Are you claiming that Google analyzes all of people's images automatically, with no consent, taken on Android and uses that data internally?

And that Amazon and GCP analyze customer info (off the disks? off unencrypted HTTP?)?

Yes, I 100% believe they analyze cloud photo backups and use the data for their own purposes, and I would be surprised if they don't explicitly allow this within their privacy policies and Terms of Use. That's the motivation for them to offer unlimited free storage. I would guess that they probably have a human look at a random image, identify the objects, and then use that to train their CV pipelines.

They do some types of automatic analysis/processing which you can see when the Google Photos app delivers an automatically composed album, panorama, or movie to you, and you can also go into Google Photos and sort your photos by the objects and/or people that Google recognizes within them.

They do this for all speech-to-text run through official Android/Google search applications. You can go to a Voice History page within your Google Account and download recordings of every "OK Google" you've performed, and I've heard that they have human transcribers who spend all day listening to random voice searches and transcribing them to provide the corpus to compare against the computerized transcription.

As for EC2/GCP, I am sure they analyze whatever data they can. I doubt they go into the disks, but they would necessarily see the endpoints of the conversations, the volume of the average transmission, and other such details. The datacenter has a lot of knowledge from metadata without having to actually introspect the bits that you're sending over the pipe or storing on disk.

Disclaimer: I am not involved with Amazon or Google in any direct way and this is mostly hearsay.

-----

EDIT: On re-reading, I realize that I flubbed in the original post (now past its edit expiration). I was in a hurry and didn't realize how flippantly my statement came off.

I don't mean to imply that Amazon/Google casually spy on the information of specific users for competitive intelligence, just that interesting metadata is available to the platform host without requiring them to look at anything that they wouldn't normally have access to, and that they already have a vast abundance of the types of images that people are uploading to public hosts like imgur.

Sure, why not?

They're already using metadata from users devices to supply traffic map data. Thought that was way cool, then discovered how it worked; Google Maps app actively reports your location behind the scenes.

Was creeped out, but I still use it all the time, despite never remembering giving explicit consent.

Be trivial for them to scrape EXIF data but not actually "peak" into the image data

I have no knowledge of what they do. I doubt the latter; that would be trouble. But the former - why not? They do it with your email.