Theoretically, they don’t have access to the photos even though they’re running the server between. That’s what end to end encryption means, it’s encrypted on your end and decrypted on the recipients end.

Unless the middle man in question retains both your and the recipients encryption keys, they can’t read the messages. This requires trusting the vendor in question.

However the only alternative to trusting a vendor is not only purchasing your own equipment, but also deploying it on your property, and building and maintaining your own isolated physical connections between those locations. This is what nation states and militaries do, the US military has an entirely physically separate and independent “internet”.

There’s a concept in information security not to spend more protecting information than that information is worth. Your gifs are probably not worth the effort of building your own infrastructure, unless you’re sending some highly sketchy content.