xAI’s new Grok feature is creating a flood of deepfakes.
"Subscribers to X Premium, which grants access to Grok, have been posting everything from Barack Obama doing cocaine to Donald Trump with a pregnant woman who (vaguely) resembles Kamala Harris to Trump and Harris pointing guns. With US elections approaching and X already under scrutiny from regulators in Europe, it’s a recipe for a new fight over the risks of generative AI."
The cat's out of the bag unfortunately. I can download stable or unstable diffusion on my home PC and make it generate all kinds of stuff. It's open source so you can't really stop that knowledge from spreading.
You can however recognize that the majority of people won't do that, and write rules around software that is delivered as a service or for a fee. That would stop 90% of it.
So while regulating GenAI is possible, it's not full fix. GenAI is kind of still the risk.
I don't care about "Grok", but just this morning I was looking at some creations that Stable Diffusion and Flux cranked out, and it's crazy just how incredible these image generators are getting.
I also saw one example of software that allows you to live-stream with someone else's face, using only a single photo of that person. You literally can't trust anything you see, read, or hear online anymore.
My webhost also put out some AI wordpress thing that basically creates posts and images for your website automatically. What's the point of the internet if everything is fake?
That's the point I'm reaching as well. The internet as a mass of disconnected sites feels dead. Any attempt to look up information goes to pages that follow the same formula and feel very AI generated even with errors and unconsistencies typical for AI. What is the point of the open net anymore? The only value I feel like I'm getting is in specific trusted platforms or sites. It's a sad state that we've reached.
I'm finding more and more websites with these long form question and answers about a topic. They all look the same and if you know enough about the subject you can easily tell that either AI or an idiot wrote these "articles.
I'm just going to start archiving human created content from the last 20 years, and stop looking for new content. The internet of 2024 is 99% quantity and 1% quality (from sites dating back more than five years).
I suspect the greatest threat is not that politicians and billionaires can create realistic images of their enemies, rather they can deny any evidence of their own misdeeds.
Harris' airport crowd serves as an example. Whether or not it really happened (I assume it did) MAGAs will have cause to plausibly believe it didn't.
Not that evidence to the contrary of belief systems has ever been effective at deprogramming those invested in their worldview.
While I share this sentiment, I think/hope the eventual conclusion will be a better relationship between more people and the truth. Maybe not for everyone, but more people than before. Truth is always more like 99.99% certain than absolute truth, and it's the collection of evidence that should inform 'truth'. The closest thing we have to achieving that is the court system (In theory).
You don't see the electric wiring in your home, yet you 'know' flipping the switch will cause electricity to create light. You 'know' there is not some other mechanism in your walls that just happens to produce the exact same result. But unless you check, you technically didn't know for sure. Someone could have swapped it out while you weren't looking, even if you built it yourself. (And even if you check, your eyes might deceive you).
With Harris' airport crowd, honestly if you weren't there, you have to trust second hand accounts. So how do you do that? One video might not say a lot, and honestly if I saw the alleged image in a vacuum I might have been suspicious of AI as well.
But here comes the context. There are many eye witness perspectives where details can be verified and corroborated. The organizer isn't an habitual liar. It happened at a time that wasn't impossible (eg. a sort of 'counter'-alibi). It happened in a place that isn't improbable (She's on the campaign trail). If true, it would require a conspiracy level of secrecy to pull of. And I could list so many more things.
Anything that could be disproven with 'It might have been AI', probably would have not stuck in court anyways. It's why you take testimony, because even though that proves nothing on it's own, if corroborated with other information it can make one situation more or less probable.
I don’t have the hope you do. The sheer number of people that believe the moon landing was faked is just plain crazy. There were soooo many people involved with that process, yet it’s still not believed.
People are and always have been divorced from reality. Some people barely understand how their own house works, but nobody ever understood everything, and at no point you could be sure you know with absolute certainty what was going on 1000 miles away from you.
Let us not act like generative AI is going to break the 4th wall of our perfect world. And I don't even mean video or even image manipulation - basic text can and has been used for manipulation forever.
Perhaps we should reexamine things we've had for decades, like signing content with PGP and building actual trust around the world.
Asking multiple times will get you variations with different policies, some of which sound distinctly un-X-ish, like “be mindful of cultural sensitivities.”
A lot of the commercially available image generators will block certain terms, like copyrighted characters or presidential candidates, or celebrities in their underwear. You can of course do this all without limits on your own hosted versions of stable diffusion and whatnot. But a commercially available option without any limits is actually news.
The reason commercial image generators put those limits in place isn't because they want to uphold an ethical use of their technology, but because they will get sued if they create a platform for making harmful images.
Musk seems to be a big fan of getting sued lately, so best of luck to him on getting even more lawsuits against him.
Interesting. Just tested this on the publicly o one I use and typed in "President Biden, White house". Sure enough it produced pictures of him standing within the White House but none of them had his face being displaying correctly. Either messed up, blurred out or just looked horrifying.
My wife talked about how Grok apparently seemed promising in AI benchmarks for a while until everyone realised the way they’re winning is by absolutely blindly outspending everyone on GPUs and brute forcing the fuck out of the problem rather than having good anything.
This seems to match that approach (and, more generally, everything Musk is doing).