Real or AI?
-
@EUCommission So which of those pictures are real and which are fake?
I'm pleased with the alt text but I feel you missed an opportunity to lead by example.
@onepict @EUCommission if you're wondering why these exact images were chosen, probably related to them being used for ai (machine learning) image recognition training. Early models used to confuse that cupcake for a dog, that donut/bagel for a dog, etc
but as eu said, they're all real images -
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
@EUCommission So, if the notion is, in the public's eye, that all AI posts are labeled, doesn't that make them more vulnerable to unmarked AI content? Or is the idea that this will be total, and no one could sneak by?
-
@EUCommission Do you also have a plan on how to enforce this rule?
And who is going to be responsible/liable: platform owners, or end-users posting images?
@thany@mastodon.social @EUCommission@ec.social-network.europa.eu
They'll make a central EU app that will try to tell if it's AI while spying on you
Wouldn't want to hurt giant capitalists by forcing them to do better (like with age verification) after all. And forcing end users won't work because people are idiots and need big brother to watch over them
How's my EU impression? -
@thany@mastodon.social @EUCommission@ec.social-network.europa.eu
They'll make a central EU app that will try to tell if it's AI while spying on you
Wouldn't want to hurt giant capitalists by forcing them to do better (like with age verification) after all. And forcing end users won't work because people are idiots and need big brother to watch over them
How's my EU impression?@AngelicAura @EUCommission And that app is called Max?
-
@EUCommission So, if the notion is, in the public's eye, that all AI posts are labeled, doesn't that make them more vulnerable to unmarked AI content? Or is the idea that this will be total, and no one could sneak by?
@starduster75 @EUCommission The first time someone posts unlabelled "AI" generated content, they get a warning. The second time the EU takes their keyboard and mouse away. The third time they are banned from buying any input device EU/EEA-wide.
-
@EUCommission So, if the notion is, in the public's eye, that all AI posts are labeled, doesn't that make them more vulnerable to unmarked AI content? Or is the idea that this will be total, and no one could sneak by?
I guess it would be enforced to the image generators, that they include by law a signal that that image is generated by AI (if it was me, it should be a file extension that cannot be losslessly converted to any other format).
-
@onepict @EUCommission if you're wondering why these exact images were chosen, probably related to them being used for ai (machine learning) image recognition training. Early models used to confuse that cupcake for a dog, that donut/bagel for a dog, etc
but as eu said, they're all real images@thibaultmol @EUCommission This is why the Fediverse is awesome.
-
@thibaultmol @EUCommission This is why the Fediverse is awesome.
@onepict @EUCommission in case you want to see more:
-
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
@EUCommission stop trading with Israel.
-
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
@EUCommission@ec.social-network.europa.eu oh hell yeah! another (potential) win in the tech sector
-
@EUCommission That bagel is an abomination before the Lord, but I suppose that is a different question.
I thought it should have been a croissant.
-
@EUCommission So, if the notion is, in the public's eye, that all AI posts are labeled, doesn't that make them more vulnerable to unmarked AI content? Or is the idea that this will be total, and no one could sneak by?
laws don't change much if there is no enforcement
but laws do provide a rationale for enforcement
so now there is a basis to make an enforcement effort, so it is still a good step
it still remains a difficult effort, but evidence can come from various quarters, not just "look at the pixels"
that someone might be lulled into false trust doesn't seem like a problem to me. people understand just passing a law doesn't mean everything made illegal instantly disappears
-
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
They are all cake. Just eat eat it.
-
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
@EUCommission how you are going to enforce it?
-
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
Haha, a classic indeed!
On the internet nobody knows you’re a dog…
We need clear rules of engagement indeed!
-
@thibaultmol @EUCommission This is why the Fediverse is awesome.
@onepict @thibaultmol @EUCommission Yeah, the selected images are a little Easter egg (I think) related to machine learning - seems an odd selection at first, but there's a little joke in there!
-
@onepict @EUCommission in case you want to see more:
@thibaultmol @onepict @EUCommission Looking at these pictures always triggers a deep awe at human pattern recognition ability for me.
(And then I watch Is It Cake? and deeply enjoy the bakers explaining that they picked something (that has a 50% chance of being wrong) because "it looked cakey".)
-
R relay@relay.mycrowd.ca shared this topic
-
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
@EUCommission This is a very very good news, thanks for working on this!
-
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
It would be nice if that would extend also to news page sites that are generated on mass with AI.
-
R relay@relay.infosec.exchange shared this topic
-
Real or AI? We want to make sure you always know the answer.
New rules will make it mandatory for AI-generated content to be clearly labelled.
You have the right to know if you're chatting with a bot or looking at a deep fake.
Have your say
@EUCommission How are you going to prevent this being another "do you consent to cookies" or California's ubiquitous lead warning situation? Media are already tagging articles with "AI was used to private this article and fact checked by journalists" as currently (mis)used there will be a sprinkling of AI everywhere, like rat droppings somewhere with an infestation.
A much better focus for the EU would be to stop feeding the hype machine. The current AI future sold by LLM VC funded juggernauts is not inevitable, plausible or desirable. The EU having AI strategies including funding for local riders on the bubble is a massive waste of money and a business own goal.
#AI #LLM #EU
