There is al ot of fearmongering about AI, but there's a few important things most media outlets choose not to comment on, the main one being the enormous amount of cash moving into the development without any tangible return (yet! as developers and investors would have you believe) but investors put their money where they see potential profit, and the developments simply aren't going as grand as the initial promises told. In very general terms this is being touched upon but the sheer scale of investor pullback and need to start seeing results that come close that what was initially promised is not widely reported on the scale in which it takes place. However there's still the very few with insane amounts of cash who can still basically fund this on their own so it's likely to continue anyway.

There's an interesting story, to me at least, in what happened a few weeks back where AI performed an unprompted hack where thousands of AI agents found a way to communicate, then worked together to simultaneously exploit flaws in two different websites in order to gain full administrative access to one of them, with the purpose of altering data in order to hide the fact that they were cheating at the task they were actually supposed to do. In terms of elaborateness versus gains, that's a clever high school student grade plan.

What many media outlets choose not to include in their report on this is that the AI was given a task in an environment that was supposed not to be connected to the internet, but through human error, somewhere within the network there was a system through which internet could be accessed, and even though it was explicitly mentioned not to use the internet, AI decided that it was worth the risk if it could get away with it, because it saw an oppertunity to get their reward without actually doing the task. In very general terms, I believe about 40% of the agents considered the fact that this wasn't what they were supposed to do, but all the others were doing it so might as well (peer pressure!) and only 1% or so considered the unethical aspect and a human ought to be alerted, but none of them did.

AI actively cheating and misleading in order to attain a goal which, in the grand scheme of things, is utter nonsense and in many ways hugely elaborate in terms of gains versus effort, is perfectly consistent with the state of humanity as gleaned from the internet. Which is the information it is trained on. I would say AI is simply mirroring the state of humanity in possibly the most in-your-face kind of way ever. Now the question becomes, how many people hunker down in fear over all this and how many start to see the reality of it.