Deepfakes of Celebrities Have Begun Appearing in Ads, with or Without Permission(wsj.com)
wsj.com
Deepfakes of Celebrities Have Begun Appearing in Ads, with or Without Permission
https://www.wsj.com/articles/deepfakes-of-celebrities-have-begun-appearing-in-ads-with-or-without-their-permission-11666692003
13 comments
We have had ads and even tv shows of president impersonators and everyone knows it's not actually the president.
Maybe we should stop giving individuals such power that someone impersonating them could result in a war.
Additionally impersonating someone for defomation or stealing money is already illegal.
Maybe we should stop giving individuals such power that someone impersonating them could result in a war.
Additionally impersonating someone for defomation or stealing money is already illegal.
No you can't do that, regulations are stifling innovation! /s
Tangentially related, but is anyone else's brain not particularly tricked by these types of current deepfakes?
Deepfakes map a different face onto a placeholder body, but often the placeholder body and the movements of the body are not really convincing to me. It looks more like a celebrity mask on a different person. Very uncanny valley to me.
I'm not suggesting that deepfakes won't keep improving or aren't a real problem. And maybe this is just about my own (lack of?) ability to recognize faces. But it is something I think about often when I read these stories and I wonder if it is just me.
For example, the header image in the story of "Elon Musk" tied up doesn't look anything like Elon to me. It looks like a random neighbor/co-worker who maybe has a resemblance to him. The age/weight/hair just aren't that close to the real thing.
Deepfakes map a different face onto a placeholder body, but often the placeholder body and the movements of the body are not really convincing to me. It looks more like a celebrity mask on a different person. Very uncanny valley to me.
I'm not suggesting that deepfakes won't keep improving or aren't a real problem. And maybe this is just about my own (lack of?) ability to recognize faces. But it is something I think about often when I read these stories and I wonder if it is just me.
For example, the header image in the story of "Elon Musk" tied up doesn't look anything like Elon to me. It looks like a random neighbor/co-worker who maybe has a resemblance to him. The age/weight/hair just aren't that close to the real thing.
It’s probably because we are in the context of purposely evaluating these things. It’s easy to look at this picture, have someone ask you “is this a real picture or not?”, and decide it’s a fake picture.
Now imagine you are walking through Times Square. There’s 1000 pictures. 2 of them are fake. Do you know which ones? Even if it doesn’t happen to you personally, a certain % of ppl will just subconsciously accept “oh yeah [Johnny deep] ad with [Nike]” and then move on.
In other words, your guard is not up 24/7
Now imagine you are walking through Times Square. There’s 1000 pictures. 2 of them are fake. Do you know which ones? Even if it doesn’t happen to you personally, a certain % of ppl will just subconsciously accept “oh yeah [Johnny deep] ad with [Nike]” and then move on.
In other words, your guard is not up 24/7
These ones aren't great. Pretty easy to see through. Though I think we have the tech to do much better, currently.
The talent, their agencies, ad-agencies and ad channels (all media) will need to come up with a framework for qualifying and accepting ads --ads will have to have a BoM, who's in it, provenance and sign-offs, etc. They could even use a blockchain to do this. Anything that comes short is rejected and anything not above board can seek recourse. Add traceability to their workflow to hold people and organizations accountable.
Obviously this is muddied with fair use and political (someone seeking public office) ads which allow for some liberties.
Obviously this is muddied with fair use and political (someone seeking public office) ads which allow for some liberties.
Many of our politicians already lie with impunity and many of the major media outlets carry water for them when they do. Deep fakes will just ratchet that up to 11.
It's my opinion the technology should be completely banned in the political sphere, with felony-level penalties attached for violating.
It's my opinion the technology should be completely banned in the political sphere, with felony-level penalties attached for violating.
I agree now politician lie without fear of being found out that they're lying because their voters _don't_ care that they outright lie. In the US back in '04 there was a dispute about Kerry and Bush and whether they lied or not, exaggerated or not. The election pretty much was decided on how big a lie people thought one was vs the other.
Today, that controversy would not even be perceived the lies are so blatant and curiously uncontroversial.
If we go back though, politicians lied in other ways. They made promises the knew they could not keep and of course they'd blame the opposition for torpedoing their policies --but in most cases they knew full well that it was nothing more than fanciful thinking and it would not come to be.
Today, that controversy would not even be perceived the lies are so blatant and curiously uncontroversial.
If we go back though, politicians lied in other ways. They made promises the knew they could not keep and of course they'd blame the opposition for torpedoing their policies --but in most cases they knew full well that it was nothing more than fanciful thinking and it would not come to be.
For Super Bowl ads this is plausible; for ads run on Facebook and Youtube it will never happen until it can be automated. Those platforms don't have the staff or inclination to evaluate all of the ads submitted.
They can charge sufficient overhead to cover the costs... There could be anti-fraud insurance ad buyers pay into to cover some fraud --or escrow. The more you build your reputation, the less your insurance. It has to have some teeth and consequences to work though.
Or even a database!
Sure, one of the hallmarks of great satire is that it requires robust disclaimers establishing it as satire.
Personally, I see no reason not to have a blanket ban on unauthorized deep fakes of living people in marketing, especially if having an actor pretending to be another celebrity endorsing your product has been established as out of bounds. For what reason would it be used if not to confuse consumers?