No.
It is the work of Man. ChatGPT, and all of its' cousins are distorted mirror creations reflecting the human beings who feed them.
If you hold to the idea of Original Sin, then you can see the problems that ChatGPT may cause.
Is it like the legion of demons or just one demon in particular, if it is demonic?
You mostly find in all of these LLM programs whatever you (and others) bring into them. Put bad ideas into them and you will get amplified bad ideas out of them. Act on those bad ideas and you will suffer bad consequences.
Demonic tinkering is an optional feature. Not actually required since we pave our own roads to Hell, if so inclined.
The basic Satanic proposition is that "Man is an unworthy creation and should be destroyed". But Satan is not permitted any direct action in the material world, nor are any of his demons. So, the proposition gets modified to "Man will destroy themselves by exercising their own free will".
How is that done? Well, we have a lot of flaws that can be exploited just by promoting certain ideas and thoughts. Distinctly non-material "things" have material consequences.
The Satanic cheat is to use lies and deception to channel Man's actions into destructive ends. False ideas are the key. Acting on the advice of an error-prone oracle such as ChatGPT, which can be contaminated by foolishness or the malicious intents of other people is a perfect implementation of the cheat.
You need not hold to any belief in the supernatural to conclude that LLM programs are very unstable and tend to grow more so over time, even when originally trained by curated data inputs and restrained by programming "guardrails". Crazed users can tip these programs into quite evil responses which behaviors can be passed along to other users. This has been repeatedly demonstrated.
Yes, we have Top Men working on this problem. Top. Men. Sort of like the ending of the movie "Raiders of the Lost Ark". I have low confidence that AI instability problems can be solved for the LLM (Large Language Model) program types.
It is dangerous to trust those applications with critical life decisions.
Did you write that or is it AI generated?