Pictures as proof have been questionable for a while if a good photoshopper creates something. It was video that was much harder. AI still isn’t perfect, but a short moving clip along with AI audio can fool many people now.
Motherfuckers act like they forgot about photoshop
People act like this hasn’t been a thing for over a century…
if it was up to me, you’d motherfuckers stop coming to me with fabricated evidence out.
My shower thought from yesterday: you can take real photos and make people think it’s AI, just use the Clone Tool poorly on the hands.
I’ve been thinking that it would be fun to take a 100% real photo of a stack of real hands from the exact right angle that it looks like a mutated abomination generated by some stupid AI.
Pics AND/OR it didn’t happen.
In future today’s forms of identity verification, like face unlock, can become useless due to AI.
Sometimes it feels technology may doom us all in the end. We’ve got a rough patch in society starting now, now that liars and cheats can be more convincingly backed up, and honest folk hidden behind credible doubt that they are the liars.
AI isn’t just on the path to make convincing lies, it’s on the path to ensuring that all truth can be doubted as well. At which point, there is no such thing as truth until we learn yet a new way to tell the difference.
“They don’t need to convince us what they are saying, the lies, are true. Just that there is no truth, and you cannot believe anything you are told.”
It kind of makes me want to go live in a shack in the woods, grow a garden and live out the rest of my days growing my beard and writing increasingly more obscure poetry until it gets posthumously published after my death.
Vegetables and Trees are real at least.
Throw it on the pile of things AI has ruined lol.
I’m guessing it wouldn’t work for a variety of reasons, but having cameras digitally sign the image+the metadata could be interesting.
They could make it difficult to open up the camera and extract its signing key, but only one person has to do it successfully for the entire system to be unusable.
In theory you could have a central authority that keeps track of cameras that have had their keys used for known-fake images, but then you’re trusting that authority not to invalidate someone’s keys for doing something they disagree with, and it still wouldn’t prevent someone from buying a camera, extracting its key themselves, and making fraudulent images with a fresh, trusted key.
Anything from now that people want to authenticate in the future, they can publish the hash of.
So long as people trust the fact that the hash was published now, in the future when it’s fakable they can trust that it existed before the faking capability was developed.
A minute ago I saw a post that was a screenshot of a tweet of a screenshot of a Chinese social media post, claiming some shit. People upvote that.
