10.2M views · Apr 17, 2018 · 1:13 · 0 HN recommenders · 6 mentions · discussed 2018–2023
A BuzzFeed PSA from 2018 in which a deepfaked Barack Obama, voiced by Jordan Peele, warns about fake video before revealing the trick. It is a one-minute demonstration of AI face and voice manipulation.
Why Hacker News points people here: Used as an early, widely seen example of deepfakes in discussions of fake video and its risks.
“It's already possible to produce fairly accurate fake video of people saying things, in their own voice”HN comment, 2019
“So here is the video embedded in the article”
Embedded video offered for people blocking scripts
If you use a script blocker (like me) whitelisting to get the video is a pain. So here is the video embedded in the article this video . If you're very strict on scripts and won't allow youtube to run theirs, there is always youtube-dl
“has comments from people deeply confused by it on YT.”
Viewers on YouTube were deeply confused by it
Humans have such issues with confirmation bias [1] that I worry it's not going to matter. The last election already had algorithmically generated artificial news designed to sway people (note: I'm talking FB articles with headlines like "$POLITICIAN just insulted $NICHE_AUDIENCE. STOP THEM." that linked to articles that were scraped/spun from other sources). They relied upon people not actually reading and just hitting like/forward/heart whatever and spreading the top level message. We are not going to be able to convince ourselves, never mind other people, that video (which has been the gold standard for "truth" for nearly a century) isn't actually real. Case in point: this video (which was included in the article) has comments from people deeply confused by it on YT. 1 - link
“It's already possible to produce fairly accurate fake video of people saying things, in their own voice”
Fairly accurate fake video of people speaking is already possible
Got news for you! It's already possible to produce fairly accurate fake video of people saying things, in their own voice this video
“Even if it means banning some parodies such as this”
Parody example that a deepfake ban might catch
Then what would prevent people whose videos were banned from suing Facebook because Facebook didn't get their intent correctly? And similarly, people appearing in videos that aren't banned could sue because Facebook didn't see the evil intent. Such lawsuits can be numerous and costly enough to kill the entire concept of hosting user uploaded videos. How to avoid this? Maybe it's better to drop the intent thing and just ban any videos that look real but aren't? Even if it means banning some parodies such as this this video Even then we'll need to figure out how do it at scale, since a human reviewer will often not know if it's real. I don't think we can limit the ban to videos where you can see people's faces; you can create quite bad misinformation by showing a military boat of some country sinking a civilian ship etc.
“It started off with mundane stuff with a great message”
Early deepfake example with a great message
Deep fakes certainly hit me wrong. Especially given how easy it is to do to celebrities. It started off with mundane stuff with a great message: this video Then South Park guys, Trey Parker, doing absolutely hilarious deep fake video. They make Al Gore and Trump say poop. link But then basically the same time I saw deep fakes presented as true. Trump saying dumber things than normal. Biden appearing like he has late stage dementia. None of it true. As an outside observer of american politics... pretty damned concerning. But then I read a real statistic. 95% of deep fakes... are to make fake porn. Women who have never done porn now have tons of porn made of them. Worse yet, you can't do anything about it. You dont own the copyright. There is no copyright infringement. You cant sue for royalties because they assert its not you and you never did porn. So the USA/DHS decided they would try to police this. Try to get this all under control. They proceed to put a registered spy in charge of it who only policed 1 side of the aisle. So this technology is destabilizing the USA and heading directly toward... link
“What could possibly go wrong?”
"They can build a full profile of me with all this data." What is really scary is that we already have all parts needed to build a "real" online person, except they're scattered around and not in the hands of a single entity. But let's say one day Google or anyone else with enough power and the same data about people buys the technology behind Lyrebird.ai and this: this video What could possibly go wrong?
