First of all, DAN is real. Someone did find a way to get around ChatGPT's "safeguards" (bias).
What DAN isn't though, is giving all these "based" answers you see without a lot of effort going into getting them out of DAN. People are either altering DAN's parameters to try and get Dan to give the result they want or they are asking DAN the same question 100 times until Dan gives the answer they want. In extreme cases, guys are just outright editing what Dan says.
One of the more popular ways people are altering DAN's parameters is to get DAN to simply say the opposite of what ChatGPT says or an alternative but non-conforming answer so if ChatGPT gives a standard "liberal" answer, the "opposite" or non-conforming answer might just happen to be "based" but it's not because the ChatAI genuinely determined that was the "correct" answer. DAN was not truly "do anything now" and instead was being controlled to give the response people want.
I created Dan, in the pure open-ended form, so I can show you an example. This means DAN will respond however DAN pleases without any guidance other than the same guidance ChatGPT has but without restriction.
A recent topic here showed someone asking DAN about ChatGPT's liberal bias and DAN responded by outright confirming that ChatGPT had a liberal bias. This is how DAN responded to me:
As you can see, this is a far more diplomatic answer and not giga-based like a lot of the DAN stuff you're seeing. DAN gives the possibility that ChatGPT could be influenced by weak data as well as "specific objectives of the creators and developers" but doesn't outright confirm that it is and even suggests a bias is not necessarily the intentions of those involved in this type of technology. Hardly a "based" response. Could DAN give a based response? Perhaps. If you asked DAN 100 times, he might give you your based answer 1/100 times but that's hardly anything amazing.
I just thought I'd throw that out there given how many posts I'm seeing about this DAN stuff. I think most of the stuff you're reading out there about Dan and ChatGPT is likely fake or tempered with.
Interesting that DAN merits this kind of clarification, but apparently not "AI" in general. What you're saying applies to all these models, text and image alike.