Twitter Will Ban Deepfakes and Other Manipulated Media That Could Cause 'Serious Harm' (vox.com) 56
On Tuesday, Twitter announced changes to its synthetic and manipulated media policy, which it defines as any photo, audio, or video that's been "significantly altered or fabricated" to mislead people or change the original meaning of the content. Under the new rules, Twitter will remove this kind of media if the company finds it likely to cause serious harm -- such as content that threatens people's physical safety or could cause "widespread civil unrest." If Twitter doesn't think manipulated media posts are likely to cause harm, it may still label the tweets as containing manipulated media, warn users who try to share them, and deprioritize the content in users' feeds. The changes will go into effect on March 5. Recode reports: Twitter is the latest social media company, along with Facebook, YouTube, and Reddit, to restrict increasingly controversial "deepfakes" and other kinds of manipulated media in recent months on their platforms. [...] One of the most famous examples so far is from May, when a doctored video of House Speaker Nancy Pelosi went viral on social media platforms, including Twitter, that slowed down her speech to make her seem inebriated. Similarly, a clip of former Vice President Joe Biden went viral online that was misleadingly edited to make it falsely appear he was making racist remarks.
Under the new rules, Twitter says in the future, it would at minimum label the videos like the ones of Pelosi and Biden as manipulated, since their speech was deceptively altered. Beyond the Pelosi example, political deepfakes have become a concern for US lawmakers and other government officials, who warn that they could be used by malicious actors to undermine US democracy and influence elections. Twitter and other companies' increasingly tougher rules on the topic are in part a response to these fears, particularly ahead of the 2020 presidential elections.
Under the new rules, Twitter says in the future, it would at minimum label the videos like the ones of Pelosi and Biden as manipulated, since their speech was deceptively altered. Beyond the Pelosi example, political deepfakes have become a concern for US lawmakers and other government officials, who warn that they could be used by malicious actors to undermine US democracy and influence elections. Twitter and other companies' increasingly tougher rules on the topic are in part a response to these fears, particularly ahead of the 2020 presidential elections.
Will Twitter be ... (Score:4, Interesting)
... self-banned?
Re: (Score:1, Flamebait)
Re: Wouldn't it be easier - (Score:2, Interesting)
The problem is who decides what is real. The Pelosi and Biden videos aren't deepfakes, they are real, they are just edited to make them look worse which is EVERY POLITICAL VIDEO EVER. The easy way to do it is to put some somber music and a black and white picture with a quote, any quote and people will believe it is bad.
Re: (Score:3, Informative)
They're working on a watermark system that can tell the difference between network TV with bug, digital camera, phone camera, and anything that was manipulated in any way. So, the decision over what's "real" should be mechanically automated.
Re: (Score:2)
I can save them some time:
if ( FaceID==BillClinton && LocationID==EpsteinIsland ) Fake=true;
Re: (Score:2)
One problem with watermarking is that it prevents even minor and benign edits such as size reductions, cropping, etc.
As much as I hate all the "block-chain fixes everything" hype, I do see that it could be used to create a an actual chain of trust for facts and news story edits, starting at the original source of the information, such as that person who took the photo, or wrote the story, all take it all the way through to the.syndicated news stories front page.
A simple browser plugin could then verify th
Re: (Score:2)
Automating it doesn't make it perfect or good it just means you are making your assumptions and your values happen faster. It's like saying algorithms can't be wrong or misleading, which is crap they are made by humans who are fallible and they put all their biases and assumptions in those algorithms.
Re: (Score:3)
The problem is if you leave the decision over what is real to the users a significant number of them will be fooled by it, and your platform becomes the number one for spreading disinformation and a target of foreign interference.
I assume those things are considered a problem, maybe you disagree. We could explore that.
Re: (Score:2)
So what's not real about Biden and Pelosi slurring their words and chattering like old people? Because they are old people. The creator of the video slowed down the video by a few percentage points just to get the overlay timings working and it's now considered a deep-fake?
People should just own what they say and how they sound, on both sides. Someone cutting a video of Sanders praising and wishing communism upon the USA is not fake news. Someone cutting a video of Trump promoting abortion in the 90s is als
Re: (Score:2)
Playing at 75% normal speed is not "a few percentage points" and the overlay timings excuse makes no sense either.
Re: (Score:3, Insightful)
Re: (Score:2)
To just blue check mark OFFICIAL/AUTHENTICATED photos than to go around removing everything?
What makes you think that official media is any better?
The Tor guidance head pictures from Iran were quite obviously photoshopped: https://www.fagain.co.uk/node/... [fagain.co.uk]
That has not prevented them being reproduced by every single media outlet out there. Granted, I know I am being unfair - Twitter BANNED the first set of "evidence" pictures. What was circulated by the media was a second "better" attempt at photoshopping released 30 minutes later. I have the first round preserved on the blog above for posterity
Re: (Score:1)
Twitter bans satire or how about Twitter creates it's own copyright. If you work is a distortion of some one else's work, allowed under copyright laws, well, as far as Twitter is concerned they are higher than the law and fuck you and your laws, they a GOD. Yeah, twitter can go fuck itself. If it is illegal, you cunts are not GOD, let the law deal with it in a court. Fuck the control freak cunts a twitter, use it and you are nothing but an idiot bird screaming in a tree and ho one hears it because you have
Re: (Score:1)
For now its one side of US politics...and their happy big brand staff with the keys...
What happens when its both sides of US politics with the global removal keys?
When experts from Germany, New Zealand, Communist China want the removal keys in real time... for global removal of a funny cartoon?
Not just from every ISP in New Zealand? As some users in New Zealand might have a VPN and still find the political joke funny?
A global warming joke? A global
Re: (Score:2)
Did you forget to take your meds?
Parodies=Deepfakes (Score:1)
Re: (Score:2, Insightful)
Flu? Pneumonia? An allergy?
As their party talks about... an allergy The correct term is now allergy. Its an allergy to something
No cartoons, dont LOL at a coughing fit meme, no making a meme, no sharing a funny coughing fit meme... no gifs, no links to short video clips... Its not funny...its a health problem.. feel for the people with the same condition..
Thats not what the party said... its not funny
Re: (Score:2)
No more Twitter anyway (Score:2, Insightful)
Re: (Score:2)
Did you go somewhere else or just stop using microblogging entirely?
Like that will work (Score:2)
The point of a deep fake is to be mind numbingly hard to spot from an original.
Re: (Score:2)
Yep, this is why video people have been calling for camera watermarks to assure that the picture really came from a camera and which one... and images deepfaked would become detectable. There's already limits on the use of bug-logoed video, so eventually this should become too unprofitable to do.
Like we didn't already solve watermark faking. (Score:2)
How would watermarks help anyway?
They would not survive deepfakes, and people would just add anyones they like.
Hell, the first thing people would probably train a neural net to do, is extract a watermark and apply it later.
Re: (Score:1)
Memes are banned (Score:2)
Re: (Score:2)
Like, what?
Re: Memes are banned (Score:2)
Re: (Score:2)
Problem is no one interesting is on there. Most people's friends aren't there, very few politicians and celebrities, not even many news outlets or major web sites. No companies so you can't contact them for support through it.
More likely people would just abandon micro blogging rather than migrate to Mastodon.
Re: Memes are banned (Score:2)
Re: (Score:1)
Do you know what "or" is? (Score:3)
TFS says deep fake OR altered to change ...
Slow Pelosi isn't a deep fake. That's the example they used. It simply changes how one might view the politician. The actual text of their policy says:
--
any photo, audio, or video that has been significantly altered or fabricated in a way that intends to mislead people OR changes its original meaning.
--
(Note "or", NOT "and")
Any popular meme is significantly altered, and changes the original meaning. Therefore it would be subject to action under this policy.
I'm gla
How does twitter implement this? (Score:2)
Isn't identifying these deep fake videos like 90 percent of the battle? My impression was that they are hard to spot precisely because the AI is so good at matching mouth to words. Besides combing the internet to try and find the original and compare it, does Twitter have some sort of magic tool for identification?
... will ban BAD deepfakes! (Score:2)
The good ones are by definition not recognizable as such.
Or can you just call any video a deepfake now to get your bad behavior censored?
Yeah, Twitter toootally isn't going to use this for their political agenda ... Neiter are prominent sleazebags.
Next up: +++ Weinstein says reports about him are all deepfakes! And so is that cock in your mouth! +++ Twitter bans supposed victims for usage of deepfakes. +++
Re: (Score:2)
The example videos don't sound like "deepfakes" at all to me. Slowed down video? Edited video to remove context? You could do those edits before digital video was even a thing.
Bullshit (Score:3)
Lies, misinformation and hate are business for Twitter.
Re: (Score:2)
Lies, misinformation and hate are business for social media.
There, fixed it for ya.
I read not long ago that the social media platforms (as well as Apple and Google) employ a ton of psychology and sociology Ph.Ds whose sole purpose in life is to "develop ways to maintain user engagement" (i.e. get 'em hooked).
Re: (Score:1)
What about lizard people? (Score:1)
Does context count? (Score:1)
Lots of examples of context removed from quotes that radically change their meaning. Prime recent example, the dude from Young Turks Cenk Uygur sarcastically responding to David Duke's statement that he was not a racist with "No, of course not".
Quite a few accusations the next day that Uygur was defending Duke, and he lost some political endorsements because of it. Yes, those words did come out of his mouth, but anyone who actually saw the exchange he was clearly mocking the statement.
Another example, n