Epistemic Status: Based on my own experience and intuitions only.
Some reasoning heuristics, that the brain uses, seem to rely on that there is limited information available. And newspapers, TV, and search engines break them hard.
Assume, that you want to evaluate if something is really good. A heuristic that your brain seems to use is to look at all the available data and search for bad things. If you reach some threshold of cumulative badness, then you will conclude that the thing can't be really good. The problem is that the threshold does not properly take into account, from how much data you select your observations from. So if you have enough data, you will always hit the threshold, so long as a bad thing occurs with a probability greater than zero.
Assume that the probability of being born a psychopath is 2%, that you live in a hunter-gatherer tribe of 12, and that you see 5 generations over your lifetime. Then you see around 60 people in your life within your tribe. You will have a 30% chance of not ever seeing any psychopath. And the psychopaths that you might see, might not be dumb enough to get caught, or might not do anything bad in the first place. So it is really unlikely that you will see any really bad behavior, caused by people being psychopaths.
Now assume you are thinking about the (horribly underspecified) question of if humans are innately good. It seems that in this situation, you would be more likely to think so because it is unlikely that you will observe a lot of bad behavior. And I would expect that the badness threshold is calibrated based on situations like this.
Compare this to the situation today. Even if we only consider video material of the last decades that is publically accessible, we have an unending fountain of examples of people who did terrible things. Just type psychopath, serial killer confession, murderer, or related terms into youtube.
Using the heuristic described in the beginning, your brain might very quickly conclude that humanity is evil if you are not careful.
Let's ask a better question. What is the percentage of human beings that do terrible things to no good end? Asking this question, it becomes much easier to see, that we need to know how many people are there in total, and how many of these people did terrible actions how often.
The problem is not only that our brain uses this heuristic. It is also that we normally ask questions in such an ambiguous way, that this heuristic seems to be applicable in the first place.
Another issue is that media, if it is curated, is prone to talk about the terrible and ignore the good. This creates a bias. Reporting on all the murderers, but not on all the people that did not do anything terrible. And there are a whole lot more of those.
A similar problem exists with search engines. If you search for information on terrible things, then you will find it. And you might not think of searching for the good things. But even if you did, it is not like you can look at all of the available information (which of course is also biased towards focusing on extreme events, and perhaps more towards the negative than the positive). You can't look at all the positive things and all the negative things and then count them and calculate some weighted ratio. From your perspective, there is just an unending amount of information pointing in either direction (at least if try to go about it in a naive way).