@skinnylatte since nobody else has said it yet: the correct thing to do in this situation is NOT to complain about the lack of alt text. That's perfectly appropriate in a number of situations but you have to judge who you're responding to. Instead, either just move on with your life, or better, post a reply with "Image ID:" and describe the image yourself. Federation does sometimes limit who sees a reply, but the OP can exit your alt text into their post of they want & have the bandwidth.

If you think the OP in a war zone should have taken the time to write alt text, then it can't really be very difficult for you sitting on your couch to do so, can it?

It's true that consistently posting without alt text shows some combination of total disregard for disabled people (yikes) and/or some kind of disability or limitation that makes it hard for someone to write alt text. Criticizing lack of alt text without considering the later possibility shows the same kind of disregard. Volunteering alt text and seeing how the OP responds is usually the best first move. If you're in the group that struggles with alt text, the #Alt4Me hash tag will not only summon a volunteer to help out, but this will also naturally boost the visibility of your post.

There are reminder bots you can follow for reminders on your own posts if you forget; I don't know of any non-opt-in reminder bots, which is good because of the contextual judgment necessary.

There are bots that will generate alt text using generative AI, these are bad for a number of reasons. I think that Blind users and others whose disability prevents them from comprehending images are justified in using image-to-text systems themselves, but these systems are flawed and come with serious risks, so others making that choice for them (especially if the text isn't labelled as AI-generated) is NOT acceptable. Given that #Alt4Me exists and promotes community, doubly so.

I'll repeat here my example of a concrete risk of using image-to-text without being able to verify the result (there are a bunch of other ethical issues but this one is most direct):

User A posts a picture of themselves and some Black friends hanging out at a bar, without alt text. The post says "hanging out with my friends."
User B replies with machine-generated alt text, without checking it carefully (vigilance is impossible long-term and anyways the effort necessary for vigilance would be better spent writing alt text yourself in almost all cases).
User C, who is Blind, reads the post and the reply, and comments "haha, I didn't know they let monkeys drink in bars"
User A blocks and reports user C for their blatant racism.

What was the machine-generated text? "a tall blonde man having drinks with some monkeys at a bar." This is an error (describing Black people as monkeys or
other apes) that happens regularly if infrequently in image-to-text systems, in part because of racism in the training data. It's mostly rare enough that vigilance is hard.

I can imagine user C using an image-to-text system to access images without alt text and accepting the risk that this happens. What's not okay is for user B to make that choice for user C, especially if they don't label their post as machine-generated so that C has no way to understand the risk.