For breaking news stories, the website might keep updating the pictures.

Lemmy presumably goes to the website, and downloads a relevant image to use as a thumbnail. Then the writers update the article images.

Typically, when links are published to link aggregator sites like lemmy or reddit or whatever, the link aggregator looks for a meta tag marked thumbnail to grab the URL for the image. It looks like this:

<meta name="thumbnail" content="https://path/to/img.jpg">

In this case, the thumbnail tag contains a url to a different picture, and the one used for this link is nowhere to be found, so I’m guessing that when OP created the link, Lemmy cached the thumbnail picture from the provided URL, and then afterwards, AP changed it in their article.

I think websites do that quite often these days. If you are trying to see the thumbnail in full resolution, they won't put it in the article, so you have to scan the entire page fruitless.

It causes me irritation.

Look up twitter cards and Facebook graph HTML tags.