I wonder how even is possible to index something like that. Is it obvious during index build that something is old? Is only needed to search ceuta 2021 - but it finds recent articles comparing current situation to 2021, not 2021 articles.
Maybe problem is not that it cannot search for those articles from 2021, but that articles from 2021 are gone now, or they are buried so deep it is so hard to index them.
Here, I can easily prove it: search for something like "site:theguardian.com" and set the date filter to show you results before say 1/1/2020. I guarantee you you won't find a single news article within search results. You will find other pages from their site such as tag pages or old job postings of theirs and what not, but no news article what so ever. Then swap The Guardian for whichever news source you want and do the same.
Then do the same on say DuckDuckGo and it'll work as intended.
Sites can tell Google what they want indexed and want not to show, and Google's crawler also is supposed to not hammer sites with unwanted indexing, and has quotas where if your site has a massive amount of content but nobody is visiting/searching for some of it, they'll limit how much they crawl/show it.
But this is something site operators want to control for their own purposes; I imagine a news media website gets very little real human traffic (and ad money) to 6 year old news articles, but a lot of bots looking for training data or just archiving their content. Considering news websites aren't loudly complaining about this I wouldn't be so sure it's not their own doing.