I don't understand why the people manipulated with lies aren't offended at the liars. They are primed to believe the lies, but aren't they mad that someone sought to manipulate them?
This keeps happening because it works. But why does it work? Why isn't there blowback?
It's embarrassing. Instead, they tell themselves, "well, it could have just as easily been real."
This has always been the case.
However, the more interesting thing going on is that we have never had to reckon with this problem anywhere near as much as we do now, since the barrier to creating fakes has been reduced from "skilled software user" to "all five billion humans who can write".
The ironic part is that the veracity of the debunking is being questioned, so depending on how this turns out, this could either be referring to "haha dumb right wingers falling for AI generated meme" or "haha dumb left wingers falling for AI generated debunking". Truly a rorschach test of a comment.
As much as I hate to have to defend Jack Posobiec (aka EndWokeness), saving the "real" screenshot from this article and uploading it to OpenAI's verification tool[0] indicates that it, in fact, was AI-generated. I welcome any correction on this.
Of course it doesn't conclusively prove that the flag was definitely right side up, but you'd really have to bend over backwards to explain why Oregon Journalism Project's "debunking" is watermarked but the counter-debunkings aren't.
There's only two images in the article, the "real" (fake) image (which appears twice), and the screenshot of the tweet; the latter returns "No OpenAI signals detected" for me, unlike the other.
Then just try looking in the bottom left corner of the "real" image to see the leftovers of the upside-down flag. The SynthID watermark isn't even necessary here to spot the obviousness of the fake.
That’s not the question. The question is “were they even there” before the “screenshot”.
As an example, I can screenshot your comment and have AI edit it to have you say “I really enjoy misleading people on the internet for money”, post it, and then say “I took this screenshot before they deleted it.”
But also just look at the bottom-left of the image as is mentioned elsewhere in this comment thread: they didn't even bother to photoshop the entire image in their "debunk".
This keeps happening because it works. But why does it work? Why isn't there blowback?
This has always been the case.
However, the more interesting thing going on is that we have never had to reckon with this problem anywhere near as much as we do now, since the barrier to creating fakes has been reduced from "skilled software user" to "all five billion humans who can write".
[0] https://openai.com/research/verify/
e: Okay, now I feel stupid for even bothering to do that. Look below her name and you'll see the leftovers of the old flag.
edit 2: Retracted. https://www.oregonjournalismproject.org/editors-note-october...
https://x.com/kgallella/status/2112332869461045733
https://x.com/Honest_Bender/status/2112301833708519659
Of course it doesn't conclusively prove that the flag was definitely right side up, but you'd really have to bend over backwards to explain why Oregon Journalism Project's "debunking" is watermarked but the counter-debunkings aren't.
And now that you’ve said that I’ve retried the same image 5 times and gotten AI 3 times and Not AI twice.
Something tells me that tools is about as reliable as Pangram.
https://www.google.com/imgres?imgurl=https://lookaside.fbsbx...
As an example, I can screenshot your comment and have AI edit it to have you say “I really enjoy misleading people on the internet for money”, post it, and then say “I took this screenshot before they deleted it.”
But also just look at the bottom-left of the image as is mentioned elsewhere in this comment thread: they didn't even bother to photoshop the entire image in their "debunk".