Papers of the day   All papers

Few-Shot Adversarial Learning of Realistic Neural Talking Head Models

Comments

Joe Rogan: They can make video of you saying things from a single photograph. I feel like the water just dramatically pulled back from the shore and we’re about to experience a tsunami of fake shit.

2780 replies, 70826 likes


Dmitry Ulyanov: Another great paper from Samsung AI lab! @egorzakharovdl et al. animate heads using only few shots of target person (or even 1 shot). Keypoints, adaptive instance norms and GANs, no 3D face modelling at all. ▶️ https://youtu.be/p1b5aiTrGzY 📝 https://arxiv.org/abs/1905.08233 https://t.co/GjVrJbejT0

953 replies, 36563 likes


Jake Clark: We are so fucked

45 replies, 3691 likes


Mark McD ☠: Samsung researchers have released a model that can generate faces in new poses from just a single image/frame (for each of face, pose). Done by building a well-trained landmark model in advance & one-shotting from that. This gif is my favourite... https://techcrunch.com/2019/05/22/mona-lisa-frown-machine-learning-brings-old-paintings-and-photos-to-life/ https://t.co/YzA9HBgUfN

40 replies, 3221 likes


Justin Whang 🐙: I can't wait to have realistic porn of my favorite centuries old paintings

48 replies, 1600 likes


Cory Barlog 🎮 #Raising Kratos: Terrifying to think of all the bad ways this will be used.

67 replies, 1477 likes


Zach Kornfeld: Y’all going too far. This is wild.

10 replies, 1458 likes


Brandon Straka: Omg. Just imagine how they’ll be able to use this to accuse Trump of things he didn’t say or do 🤦🏻‍♂️

144 replies, 1228 likes


Jeremiah Grossman: Seeing is believing. Video or it didn’t happen. In the very near future, we won’t say or believe those things anymore.

31 replies, 1104 likes


Laura Ingraham: Cool. I’m sure it will only be used for our benefit and entertainment!

151 replies, 875 likes


Jesse Averna: This is nuts

33 replies, 872 likes


cowboy online: I can't wait to confess to crimes I didn't commit 😃

8 replies, 699 likes


Massimo: Samsung AI lab develops tech that can animate highly realistic heads using only a few - or in some cases - only one starter image https://buff.ly/2HM2f5W https://t.co/Ak6WXZoYIH

24 replies, 676 likes


Shannon Morse: Well, we're fucked.

24 replies, 465 likes


VoidBurger 🍔🎮: We need more mandatory ethics classes in school

11 replies, 454 likes


Dylan Horrocks: Buckle up, people. It's gonna get weird.

9 replies, 403 likes


Robby Starbuck: Prediction: A shocking video of Trump will come out before the 2020 election using similar tech and CNN will say "although some researchers say the video is a fake, others have said it is inconclusive whether the video is authentic." Media will support this "inconclusive" smear. https://twitter.com/DmitryUlyanovML/status/1131155659305705472

37 replies, 393 likes


Cʜʀɪsᴛᴏᴘʜᴇʀ Gʀᴀʏsᴏɴ 🔜 #AWE2019: @bruces Generated from one still… https://t.co/dsDhL8vUY5

7 replies, 373 likes


ΜΔDΞRΔS: #CyberpunkisNow Another look at Few-Shot Adversarial Learning of Realistic Neural Talking Head Models as implemented by Samsung AI lab This technology can animate highly realistic heads using as little as one starter image. More information here https://arxiv.org/abs/1905.08233 https://t.co/8VuPY4OCuC

11 replies, 282 likes


Android 21 Fan Account: the next 10 years are going to be terrifying and ridiculous but I'm down https://twitter.com/DmitryUlyanovML/status/1131155659305705472

5 replies, 274 likes


🥧Enclosed Grand Dad | MARIO MAKER 2!: This has a lot of scary implications. Imagine someone using this to pretend to be someone else.

19 replies, 256 likes


Gaijinhunter: I’m usually an optimist but advances in tech like this just make me fear how bad misinformation will spread even more :/

13 replies, 191 likes


Martina Markota: The future will be crazy.. https://twitter.com/DmitryUlyanovML/status/1131155659305705472

24 replies, 164 likes


Scott Fisher: We are so fucked and I don’t say that lightly. Misinformation and fake news already run rampant without deep fakes and this tech being the norm. The future is terrifying.

4 replies, 147 likes


roadrunner01: Few-Shot Adversarial Learning of Realistic Neural Talking Head Models pdf: https://arxiv.org/pdf/1905.08233.pdf abs: https://arxiv.org/abs/1905.08233 https://t.co/31GsbQtBy3

6 replies, 146 likes


🐺 Joakim ‘konjak’ Sandberg: Again, this shit is not impressive, it's terrifying https://twitter.com/DmitryUlyanovML/status/1131155659305705472

16 replies, 145 likes


Ethan Marcotte: this is probably [piles all the world’s computers together, draws a ring of salt around them] fine, just totally [sets computers on fire] fine https://twitter.com/DmitryUlyanovML/status/1131155659305705472

8 replies, 142 likes


Joanna Rutkowska: Many people find this scary. But longer-term it might be liberating, actually. Soon the use of video to blackmail people might simply loose ground. Which might be a godsend in our camera-everywhere world.

10 replies, 135 likes


Nick Kapur: Animating photographs and even paintings from even just a single source image 😮😮😮 Deepfakes are getting scary good. We need to start doubting video as much as we doubt still images.

2 replies, 134 likes


Sheril Kirshenbaum: This is wild. And dangerous.

9 replies, 132 likes


Ancilla: This creeps me out almost as much as the boston dynamics robots

21 replies, 121 likes


Steve 인창 Ahn: One day, we won't need to model characters any more.. A.I will take over.

4 replies, 114 likes


Pascal Bornet: #MachineLearning brings old pictures to life https://arxiv.org/abs/1905.08233 @alvinfoo @FrRonconi @andi_staub @Paula_Piccard @psb_dc @ronald_vanloon @jblefevre60 @evankirstel @mvollmer1 @HeinzvHoenen @samiranghosh @YuHelenYu @MHiesboeck @andy_lucerne @ipfconline1 @spirosmargaris https://t.co/RmYJRHLwSc

1 replies, 113 likes


🏴‍☠️ Maddox 🏴‍☠️: Good thing we're all switching to face recognition for use with our credentials.

4 replies, 93 likes


Linzb0t: i'm scared of this... idky i immediately think it will be used for evil ??

9 replies, 93 likes


Zara Rahman: 1. in the *best* possible case, what do the researchers behind this think it will be used for? 2. have they at all considered the (very likely and most probable) worst case that this will be used for? 3. what are they thinking?!?!?

5 replies, 92 likes


Wolfgang Blau: It won‘t even require a video of you to make you appear having said things you have never said. Which j-school is preparing stiudents for the new verification skills required? Which news orgs, other than large public broadcasters or CNN-types can afford such verification units?

4 replies, 91 likes


Christine M. Payne: Impressive results from Samsung AI!

0 replies, 89 likes


DrawKill: you kno how cool this is gonna be when you can use it on your oc's tho-

1 replies, 89 likes


🌸 Alice 🌸: Too creepy 😫 I wonder what else can be done that isn't available to the public 👀🕵🏾‍♀️

6 replies, 83 likes


Lê Nguyên Hoang (Science4All): One-shot learning!! 2 years ago, this would have seemed nearly impossible to me...

1 replies, 82 likes


Patrick Gray: This is very cool, but also a bit terrifying, especially if your image and voice are everywhere.

4 replies, 82 likes


theMFK: we are fucked so bad

1 replies, 78 likes


emily Ⓥ: Adding this to the “technically super cool but has terrifying implications” pile which keeps growing bigger. https://twitter.com/dmitryulyanovml/status/1131155659305705472

1 replies, 78 likes


Colum Slevin: Well, ok then. There it is. It’s all over.

7 replies, 76 likes


Temp: haha i can't wait for video evidence and testimonies to be completely useless in the court of law haha :')

2 replies, 72 likes


ΜΔDΞRΔS: #CyberpunkisNow Mona Lisa Overdrive - Few-Shot Adversarial Learning of Realistic Neural Talking Head Models implemented by Samsung AI lab using Mona Lisa This tech can animate highly realistic heads using one or more images. Info: https://arxiv.org/abs/1905.08233 https://www.youtube.com/watch?v=p1b5aiTrGzY https://t.co/6xgJFkZgC1

2 replies, 66 likes


Daniel Gross: Imagine the Apollo 11 documentary, but... for everything. Einstein. Lincoln. Plato. All restored to look like a Chris Nolan film.

2 replies, 65 likes


santi: reality keeps blurring.

2 replies, 57 likes


Yury Malkov: A work by colleagues from the lab. Photorealistic avatar from a single image. Can drive Mona Lisa: https://www.youtube.com/watch?v=p1b5aiTrGzY

2 replies, 55 likes


Aaron Clauset: The AI apocalypse is here already, it’s just not the apocalypse we expected 👇 https://twitter.com/DmitryUlyanovML/status/1131155659305705472

2 replies, 54 likes


Bärí A. Williams: There's no way this technology could be used for nefarious purposes... not at all...

4 replies, 53 likes


ANSHUL KUNDAJE: Wow.

5 replies, 52 likes


Cassandra Khaw: This is not freaking me out at all

4 replies, 51 likes


Ray Rod: AI is changing the game. This is mind blowing.

2 replies, 51 likes


Yoni Freedhoff, MD: This isn't terrifying at all

3 replies, 51 likes


Kris Williams: Creepy as f#%k. "Target person" says it all. Sometimes tech goes too far.

9 replies, 50 likes


DataScienceNigeria: Samsung Deepfake AI can fabricate a video clip of you from a single photo, as AI animates photos without 3D modeling. Cautious! AI can make you appear to do or say things you never did. Watch: http://youtu.be/p1b5aiTrGzY Paper: http://arxiv.org/pdf/1905.08233… Kudos @egorzakharovdl etal https://t.co/UzyZG5Qtja

0 replies, 48 likes


Just Jeff 🇺🇸🇺🇸: To me this is scary as hell, as if we don't already have enough FAKE stories. Can you imagine how this could be used for evil instead of good?

6 replies, 47 likes


Ursula Dorada: Technology is so exciting and scary

1 replies, 46 likes


Mardial: Black Mirror season 6

2 replies, 44 likes


Stephen Floor: This is amazing technology. Welcome to the future of disinformation https://twitter.com/dmitryulyanovml/status/1131155659305705472

2 replies, 42 likes


Jon Freeman: 😮 Lord help us #GANs

1 replies, 42 likes


Pep: And here we have one of the many reasons computer science and engineering curriculums need at least one ethics course.

2 replies, 41 likes


Alexander Verbeek 🌍: 📷 This is both fascinating as well as creepy. What are your feelings about this technology? #AI #tech #innovation #art #photography

27 replies, 41 likes


Lea Kissner: I am sure no one is going to use this for anything creepy or abusive in any way. /me returns nose to "respectful systems" grindstone

4 replies, 40 likes


Ryan Brown 🎮: I have never been so terrified and intrigued by the future at the same time.

4 replies, 38 likes


Samantha Cole: a user study by researchers earned their fine-tuned model (which used between 8-32 photos, still incredibly low) a perfect score for realism https://t.co/gui2sgqe79

1 replies, 37 likes


the Game Chasers: Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should

7 replies, 37 likes


Bill Weir: This is horrifying. In the wrong hands it could destroy reputations, start wars and end truth as we know it And for what? Marilyn Monroe rap battles the Mona Lisa in some stupid commercial? At least gene-editing tech has some life-saving benefits inside its Pandora’s box

6 replies, 36 likes


Mathew Ingram: Deepfake technology is evolving

1 replies, 36 likes


Jason Buttrill: Insane. Imagine this tech during an election year.

8 replies, 36 likes


Ryan Duffin: The future is gonna be weird.

5 replies, 34 likes


Doomsday.exe: can't wait to get my hands on this and make heavy talk about how much he loves topping his Doktor

1 replies, 34 likes


t0kes: this tech will only be used for deception but the fools still don’t see it lmao LMAO

8 replies, 33 likes


Dovey Wan 🗝 🦖: I’ve dabbled in AI for last few years, tho amazed by it every time, especially the recent DL breakthrough in CV, an AI armed society is relatively depressing and dystopian future to me A crypto-powered society is much more aspiring, with individual gaining control and power

2 replies, 32 likes


Peter W. Singer: #LikeWar book chapter 8 lesson: manipulating imagery and driving it viral will not just change the future, but also the past...

0 replies, 31 likes


Motherboard: It's getting way too easy to create fake videos of people's faces: https://bit.ly/2K4QtWW https://t.co/kQVFUyDpSn

1 replies, 31 likes


Kevin Fox 🦊: Reality is doomed.

3 replies, 31 likes


Danisha Carter: told y’all AI is about to get very very scary. soon the government will have video footage of you admitting to crimes you didn’t commit, that they’ve created from one facebook photo of you. #ai

1 replies, 31 likes


Amir Sariaslan: Wow!

1 replies, 30 likes


Michael Dempsey: I wrote in my newsletter this week about how small dataset ML is going to become increasingly popular. Great example here.

2 replies, 29 likes


it's B! Cavello 🐝: Wow! This is intense. Folks were concerned about GPT-2, I guess, but one shot video GANs seem way freakier to me? The paintings feel magical, though...

1 replies, 29 likes


Mark McD ☠: @iglesiasp_ @ciberado Great question! It looks like it's 'squishing' the source frame a bit, though I imagine it would have picked up this generalisation skill from the training data used on the meta-learner. The paper has some neat examples too: https://arxiv.org/pdf/1905.08233.pdf (e.g. page 7, the 3rd row ex.)

2 replies, 29 likes


dabacon: Believe nothing

0 replies, 29 likes


Enrico Coiera: While there are plenty of reasons to worry about this technology and its potential to create ever more realistic fakery, how wonderful to see Mona Lisa talk, or imagine in the future a face surgically reconstructed from just a few selfies.

1 replies, 28 likes


dione: aight im about to head out

1 replies, 28 likes


Jonathan A. Michaels: This is creepy AF. Go watch the whole video.

0 replies, 28 likes


Brett Winton: We are still in the earliest stages with neural nets.

2 replies, 28 likes


Sören Meyer-Eppler: Not creepy at all. Not even a little bit.

0 replies, 27 likes


Adam Frank: Oh yippy. I’m sure this technology will only be used for good.

0 replies, 27 likes


Andrei Taraschuk: One day Dostoyevsky will have his own YouTube channel where he will read his works...

2 replies, 26 likes


Matthew Kobach: This issue feels bigger than privacy to me. Fake videos on social media will soon be the norm.

3 replies, 25 likes


Seth Andrews: Jesus. It’s going to a tsunami of doctored videos.

5 replies, 25 likes


Rogue 👨‍🏫: AI, solving only the most important problems first...

1 replies, 25 likes


Thijs Dhollander: Holy cr*p, this is insanely impressive. 😮😲🤯

0 replies, 24 likes


Mike: We are not prepared morally, ethically, or legally for this technology and yet they continue to push the #Transhumanism agenda 😕

1 replies, 23 likes


Four P. Jimothy: The cool thing about this is I can now falsify incriminating video evidence of people I don’t like

0 replies, 23 likes


Sarah Mackie: Well this is slightly terrifying.

8 replies, 21 likes


𝐌𝐨𝐡𝐚𝐦𝐞𝐝 𝐆𝐡𝐢𝐥𝐚𝐧 | محمـــد غيـــلان: Another reason not to follow the news or believe what you come across online, even when you “see” it with your own eyes.

0 replies, 21 likes


Ted Neward: This has frightening implications, and in time, may turn out to be the nail in the coffin of credible journalism.

2 replies, 21 likes


Seb Brixey-Williams: This is extraordinary and terrifying science. If we think #fakenews is bad today, just give it five to ten years for this stuff to become widely available. Have there been discussions about the control of this technology?

5 replies, 21 likes


Krish Ashok: Tyrion Lannister’s theory that those with the best stories will be in power just got a shot in the arm from deep learning. Private facts and private stories for everyone!

4 replies, 21 likes


Stephen Turner: Wow. Unbelievable, mesmerizing, and frightening.

0 replies, 21 likes


Matthew Ocko: Some intensive focus on “defense A.I.” to detect this is going to be really important Imagine inflammatory & highly persuasive “home videos” of public figures ginned up from a single grainy 70s photograph 😬 https://twitter.com/dmitryulyanovml/status/1131155659305705472

5 replies, 20 likes


Elgin-Skye: Fascinated by this tech but deeply troubled by what it could be used for. Not just political disinformation, I keep thinking of how it could be used against individuals - women in particular (revenge porn/blackmail). I'd love to hear @JulieSLalonde or @SuzieMDunn thoughts on this

3 replies, 20 likes


danny: Samsung’s AI teams have come up with a model for mapping facial movement to photos, and even paintings, to animate them https://arxiv.org/abs/1905.08233 https://t.co/i8cCKYUPaD

2 replies, 20 likes


boubi noubi: Can't wait to make my favourite historical figures say the n-word

0 replies, 20 likes


☢ nuke.net ☢ [R3-F0LL0W]: Fucking tech demons

1 replies, 20 likes


Mustapha Itani: 1984 here we come.

0 replies, 20 likes


Pawel Pachniewski: AI animates faces from a single or few images. Paper: https://arxiv.org/abs/1905.08233 https://t.co/NSsv73TNAH

1 replies, 19 likes


Sara 🐍: COOL

1 replies, 19 likes


Dan Kaminsky: Like I keep saying, AI is compression. Lossy compression and the loss function refer to the exact same error minimization. Video codecs over the next few years are going to get quite good, especially for teleconferencing. Small vectors here. VR/AR is not just fun and games.

3 replies, 19 likes


Sundae Tecnica International: good to see the future of identity theft is strong

0 replies, 19 likes


Henry Ajder: One of the current barriers to making realistic #deepfakes is access to large, high quality sets of training data. This research from Samsung AI suggests this won't be the case for long.

0 replies, 19 likes


Emily H E X A Crose: This technology is incredible. The one thing that’s always interesting to me are the parts the intelligence has trouble with. Rendering consistent clothes for instance is really difficult if you’ve only trained the neural network on faces.

1 replies, 18 likes


ilkan: Of Off Offf

1 replies, 18 likes


JayMan: Now that’s scary

0 replies, 18 likes


Vinit Masram: Be ready for shit load of fake stuff and a great bunch of resulting lawsuits. #AI #Science #Future

1 replies, 18 likes


Jomny Wic: nightmare

1 replies, 18 likes


Darshan H Sheth ✨ @Iamdarshan: #MachineLearning brings old pictures to life https://arxiv.org/abs/1905.08233 Via @pascal_bornet cc @darshan_h_sheth @Paula_Piccard @psb_dc @ronald_vanloon @jblefevre60 @evankirstel @mvollmer1 @HeinzvHoenen @samiranghosh @YuHelenYu @MHiesboeck @andy_lucerne @ipfconline1 @SpirosMargaris https://t.co/F8f8OUQ9PE

0 replies, 18 likes


The Good Doctor: This is what they're rolling out, imagine the high level shit they still have behind the scenes.

1 replies, 17 likes


Aedonis Bravo: how long before y'all think they take that picture of Michael Jordan & actually make it say "fuck dem kids"?

3 replies, 17 likes


Mike Mozart: Fantastic and terrifying

0 replies, 17 likes


Yanick Fratantonio 🌴: Terrifying...

0 replies, 17 likes


Max: Photographic evidence finally useless

2 replies, 16 likes


eating raoul vaneigem: Arbitrarily deciding to be a deep fake accelerationist and will start incorporating glitchy machine learning artifacts into my facial expressions. Wanna destroy any recognizable distinction between artifice and reality.

0 replies, 16 likes


Maria Chong: Oh wowwwww.

3 replies, 16 likes


Shambhav Sharma: Oh this is not gonna be good

1 replies, 16 likes


Alvaro Videla: This is why we didn’t deserve Susan Sontag nor John Berger.

4 replies, 16 likes


💖Taudry Hepburn💖: First we taught sand how to think. Oh how clever it would be, it could do our numbers for us, what fun. Then, for some goddam reason, we taught it necromancy, and just expect everything will turn out okay somehow?!?

2 replies, 16 likes


Victor Dibia: Video generation (for faces) from single images is almost practical.

1 replies, 16 likes


Omri Ceren: Oh good.

1 replies, 15 likes


Alex Libertas: This is an absolutely awful thing to create. AI that can create an animated image of a still photo. Combine this with AI powered voice modulation and this opens up all sorts of issues with fake media being produced No good will come of this

4 replies, 15 likes


Knowing Neurons: Is this the future? Artificial intelligence inspired by the architecture of the brain learns to animate static faces.

0 replies, 15 likes


Fr. Leo Patalinghug: This is interesting, cool and frightening. When technology advances but morality and consciousness doesn’t, I can predict this technology will be abused by many bad people to do very bad things. I pray parents learn how to talk to kids about this stuff ASAP!

1 replies, 15 likes


orei: holy shit college socialist guy could use this to bring vladimir lenin back to life

0 replies, 15 likes


Shawn Coss: I'm totally gonna make Jesus say "eat me" it'll get you into heaven

0 replies, 15 likes


Kim Crawley: You won't be able to trust video anymore. #cybersecurity #AI

1 replies, 15 likes


Thomas en Vrac: 🤨 Flip...pant...

0 replies, 15 likes


David Price: Here is the end of “truth” on the internet ... no future video will be trustworthy. You will never be able to believe what you see. A great example of the negative impact of AI...

1 replies, 14 likes


Joris Meys: That's impressive, exciting and scary at the same time.

1 replies, 14 likes


Sandra: Scary stuff … you won’t believe your own eyes much longer.

2 replies, 14 likes


hesaitix: all the klimt women were scary/hot viennese gallerists all along?? https://twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 14 likes


Osiris Media: This is...terrifying.

1 replies, 14 likes


anarch de droite: The next stage of the virtual onslaught is in preparation.

0 replies, 14 likes


Kidmograph: WoW

0 replies, 14 likes


Bullneck: Pretty cool how everyone working on deepfake technology has no moral core nor any ability to perceive how their technology will likely destroy the ability for people to communicate or receive news in a trustworthy manner in 10 years or so.

3 replies, 13 likes


_D: we don't even need this

3 replies, 13 likes


Alessandro Nai: OK, this is getting a bit scary. How long before videos (oh hi, Strache) are completely dismissed as "yet another fake"?

2 replies, 13 likes


OLIMEX Ltd: Wow the technology can cheat us now.

1 replies, 13 likes


Tony Roberts: How long before this deep fake animation technology is weaponised for political ends?

2 replies, 13 likes


Thijs Janzen: It doesn't get much more scary than this.

0 replies, 13 likes


KatLab: Frightening...'animate' anyone, saying anything.

1 replies, 13 likes


yeehaw: Yikes

4 replies, 12 likes


Vijayendra Mohanty: Fake news is going to get worse. Solution? Wait before you jump to conclusions. Give things time to settle. Don't feel obliged to have an opinion when you have insufficient data. Feel okay with saying, "I don't know yet."

3 replies, 12 likes


April C. Wright: The #AI for #deeplearning output gets better while #detection lags I discuss the potential societal implications - both terrifying and positive - of #deepfakes (so far) at these cons: @OReillyMedia #VelocityConf https://conferences.oreilly.com/velocity/vl-ca/public/schedule/detail/74823 @CircleCityCon https://circlecitycon.com/talks/deepfakes_if_anything_can_be_real_then_nothing_is_real/

0 replies, 12 likes


KONGDONG: Never believe anything on television

1 replies, 12 likes


Damian L. Pulcini (Mr. 2008) #UTR: And my friends, let the games {officially} begin.

5 replies, 12 likes


内澤旬子/CF返礼品、粛々と準備中: !!

0 replies, 12 likes


Estela Navarrete: Reality does not exist anymore. 😮 #AI

0 replies, 12 likes


Stanley Pignal: This feels like a window into what our future will look like. Forget "deep fakes", it will go much much further-- some good, some bad.

2 replies, 11 likes


Evan DeSimone: Good thing everyone spent the last ten years taking a billion selfies. https://twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 11 likes


🇮🇹Ginz🇮🇹: never take a picture of your face again.

3 replies, 11 likes


Satomi Hiyama: New technology enables to animate face images as if they were lively talking, from only few (even a single) pictures. I wonder how it looks if one applies this technology for the middle-aged Buddhist art... 🧐

0 replies, 11 likes


🇺🇸 Drybones🗽ム: https://t.co/2vEHtQhcKi

2 replies, 11 likes


BigData IoT BI: Amazing use of #technology. I see we don’t need to worry about the #entertainment #industry for now. #AI #Animate #Movies #Portrait #Samsung #Artificialntelligence

0 replies, 11 likes


Bilal Zuberi: We are soon going to start seeing tons of content that will be hard to believe. And it will be wise of us to not believe everything we see. AI is fast transforming creative spaces. At Lux we're investing in generative AI, as well as technologies that help explain AI.

1 replies, 10 likes


Renee Yao: Impressive work using adversarial learning of realistic #NN to create personalized talking head model. Pretty cool to see Mona Lisa now has a living portrait. #deeplearning #AI YouTube: https://www.youtube.com/watch?v=p1b5aiTrGzY&feature=youtu.be Paper: https://arxiv.org/abs/1905.08233 https://t.co/ncarjhtWMb

1 replies, 10 likes


Nelson M. Rosario: If it wasn't already, "question reality" it about to become the default rule.

3 replies, 10 likes


Aarti Mann: Am I the only one who finds this terrifying?

1 replies, 10 likes


[Nick]⚡️[BTC - ŁTC]: Scary stuff. This is where I think digital cryptographic signatures will be needed. Digitally sign any legitimate video, and unsigned footage should always be seen as suspicious

0 replies, 10 likes


Mike Hind: Coming to elections near you, soon, the perfect disinformation tool

1 replies, 10 likes


Abigail Brady: Holy fucking shit. Just look at this gif everyone, look at this gif.

2 replies, 10 likes


Wetterschneider: This is part of the tech we're going to use to replace actors in films on the fly.

1 replies, 10 likes


Hans Kristensen: Good thing humans will have enough self-control and dignity not to misuse this technology! Oh wait...

1 replies, 10 likes


Gareth Lancaster: Wow. Equal parts amazing and scary in its potential.

0 replies, 10 likes


Cʜʀɪsᴛᴏᴘʜᴇʀ Gʀᴀʏsᴏɴ 🔜 #AWE2019: @bruces We're in trouble… A convolutional neural network (#AI) can generate realistic/believable video from a single still image. This is the work of a Cornell team: https://arxiv.org/abs/1905.08233 …as was shared by Bruce, to which I replied with the animated sample (and his ReTweet took off).

0 replies, 10 likes


Maarten Keulemans: Holy fucking fuck!

0 replies, 10 likes


Blender Sushi Guy: This kind of tech in a wrong hand 🤖 helps spreading fake news... 😔

2 replies, 10 likes


Prashant: They can make video of you saying things from a single photograph. I think we're gonna face some serious fake news in future.

0 replies, 9 likes


Dr Mark D'Arcy: You think we had fake news before. You ain't seen nothing yet!

0 replies, 9 likes


Pepe Jackson: @ByronFackenthal @mizdonna @travlnjak @realAuntieVodak @BluesBrother91 @LouiseT80216 @SpicyNoodles2 @Torian_Girl @LJT_is_me @a1semp @WilhelmDerErste @AllanKirkhart @weebutmightiest @ModBond1 @HankMullaney @BasedPompeo @birockin @politicalwilli @CalThread @Cara_TXZEAL @wmahoney5 @sportsmom9433 @magarnets @305local @Scrambu11 @UrUnpaidPundit @jskielb3 @nhdjinn @Noncon49285 @BS2KZ @almostjingo @RadGeekpartduex @BabeReflex_8 @blc3428 @salty_dawg69 @Debbie22210 @olystubbies44 @Kekwillsit1 @passion_ii @NHMommy1 @UniverseatLarge @unseen1_unseen @sc_lorraine @ljcljf @Drpainsinfo @SonofMan__2 @coco_bean88 https://twitter.com/dmitryulyanovml/status/1131155659305705472?s=21

6 replies, 9 likes


Ryan Stolier: Next gen fake news will end the world.

0 replies, 9 likes


Soraya Jaber: Waw amazing and terrifying at the same time 😍😱

0 replies, 9 likes


Andy Müller: animate heads using only one of few shots of target person @ #Samsung #AI @ipfconline1 @fklivestolearn @antgrasso @guzmand @AntonioSelas @leimer @sbmeunier @DuenaBlomstrom @andi_staub @helene_wpli ▶️ http://youtu.be/p1b5aiTrGzY 📝 http://arxiv.org/abs/1905.08233 https://t.co/MKPJ5SgTj8

0 replies, 9 likes


AKAI Digital: Researchers at Samsung AI have created a new system able to generate a video of a person speaking and making ordinary expressions, from just ONE single image of this person’s face. https://arxiv.org/abs/1905.08233 #artificialintelligence #deeplearning #machinelearning #NeuralNetworks https://t.co/4hcKgR1R4v

0 replies, 9 likes


Pascual de Juan: Doom is near!

1 replies, 9 likes


Grace Loh Prasad: This is terrifying 😱

1 replies, 9 likes


Compassionate Capitalist: @LisaMei62 Exactly. I find it interesting that they are spreading this tweet which Obama is talking about. Don't be fooled. He knows his activities are Treason. https://twitter.com/DmitryUlyanovML/status/1131155659305705472?s=19

1 replies, 9 likes


Henning Sanden: AI is turning scary and awesome. I can't wait to have a commercials CG AI tools available!

0 replies, 9 likes


Kraken: These living paintings are so fascinating!!

2 replies, 9 likes


Jason Caffoe @TCAF!: This is, hands down, one of the most terrifying things I’ve ever seen.

0 replies, 8 likes


BC: Bringing still images and paintings to life using #MachineLearning of realistic neural talking head models (by Samsung AI lab @egorzakharovdl et al) e.g. #MonaLisa https://youtu.be/p1b5aiTrGzY https://arxiv.org/abs/1905.08233 https://t.co/spSjsPKiWG

0 replies, 8 likes


💀Chris Priestman💀: This is cool but oh lord the tech is so gonna be used by bad people to spread lies https://twitter.com/DmitryUlyanovML/status/1131155659305705472?s=19

2 replies, 8 likes


Juan Ortiz Freuler: some argued deep fakes were not as problematic as initially assumed. Their proof: it wasnt used effectively in any recent elections. This could change: tech advances + internet penetration into areas where people have less exposure to traditional media+depend on zero rating)

1 replies, 8 likes


Chris Bambrough: So creating animated facial images from a still photo of a face is now do-able. In a year or so it'll be indistinguishable from the real thing. This is going to make telling truth from lies on film more difficult, isn't it? Anyone else a bit worried? https://twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 8 likes


Lukasz Olejnik: Photorealistic video production ("talking heads") just got easier. Only a few pictures needed. You can make Einstein or Mona Lisa alive. Sometimes only one picture is sufficient. Ideas of misuse? ;-) https://arxiv.org/pdf/1905.08233.pdf https://t.co/zhnnmuY8md

0 replies, 8 likes


Hué: Imagine the porn industry gets involved with this?

3 replies, 8 likes


Linkabel: This is cool, but scary.

0 replies, 8 likes


Trump Derangement Syndrome Therapy Inc.: Real News Will Become Even More Irrelevant As There Is No Discerning Reality. Like CNN Today.

1 replies, 8 likes


four: This is crazy and well worth watching.

1 replies, 8 likes


Taras Kovalenko / 赤井福竜: Researchers at Samsung AI have created a new system able to generate a video of a person speaking and making ordinary expressions, from just ONE single image of this person’s face. https://arxiv.org/abs/1905.08233 #artificialintelligence #deeplearning #machinelearning #NeuralNetworks https://t.co/H3vsNygwyg

0 replies, 8 likes


Dr Nadira Faber: Amazing! ... No, terrifying! ... Amazing! ... 🤯 ... Probably both. #AI #ArtificialIntelligence

1 replies, 8 likes


Joshua Livestro: What could go wrong?

2 replies, 8 likes


Alena Vranova: ~ It's my face that you see but you know it may not be me. ~ #MyDataMyLife

0 replies, 8 likes


Maja Bäckvall: I find this both unsettling and cool.

0 replies, 8 likes


MUGATU: Not impressed. Hogwarts had this tech hundreds of years ago.

0 replies, 8 likes


Glen Johnson: Totally not creepy https://twitter.com/DmitryUlyanovML/status/1131155659305705472

1 replies, 8 likes


Allikuja @ sleepy moo🌻❤️🐶💣: Can we have Harry Potter newspapers now?

0 replies, 8 likes


Fecal Sorcerer Supreme: Oh fuck this

2 replies, 8 likes


Meowski Catovitch: I wonder if this can/will be used with video compression. If it can be done in real time, the decoder could just be told "build a face our of this frame, then rotate it to these 3D positions for the next 5 frames".

1 replies, 8 likes


Jay W. Richards: Wow. Realistic video produced from a single still photograph.

3 replies, 8 likes


Mr. Santos - The Return: Somethings shouldn’t be invented, this is one of them.

1 replies, 8 likes


Hippie eMom Eevi🌲STOP USING PLASTIC: I truly hate this

1 replies, 8 likes


Mohit Thatte: Big win for fake news.

0 replies, 8 likes


Jen Gomes 🇨🇦🇵🇹: And we think fake news is a problem now? We ain't seen nothin' yet..#terrifying

2 replies, 8 likes


Ben Harris-Roxas: Stop taking selfies? (though given the limited number of shots required to generate the model I suspect that won’t make a difference)

3 replies, 7 likes


dolly: We are not ready for the deepfake era

0 replies, 7 likes


FKA SpokenELLE: Keep fuckin with these snapchat filters, meanwhile the government creating a fake you to keep on file for later 😂😭

0 replies, 7 likes


drew olanoff: absolutely frightening.

0 replies, 7 likes


Fun Machine Learning: Talking head model!

0 replies, 7 likes


Tom Hartley: More neutral network magic. Talking heads from a few photos or even paintings (a la Harry Potter)

0 replies, 7 likes


JoukoSalonen: wow

0 replies, 7 likes


Vanessa (Berben) Buck: Welp, this is terrifying...

1 replies, 7 likes


Falgun H. Chokshi, MD, MBA: This is amazing. And pretty freaky. #AI #GAN

1 replies, 7 likes


Lodewijk van Zwieten: Next level deepfakes incoming...

0 replies, 7 likes


Paul Minda: I also thought The future would resemble Star Trek, but it turns out it resembles Harry Potter

1 replies, 7 likes


Lamhfada: This is going to be creepy in the future...

2 replies, 7 likes


Daniel W. Dieterle: Scary tech...

2 replies, 7 likes


TechAnalyst: This is scary stuff. Be prepared the the coming era of #deepfake 😨

0 replies, 6 likes


Björn Friðgeir Björnsson: I foresee no problems with this technology

0 replies, 6 likes


Polynomial - C: #AI #Dostoevsky #UncannyValley #PatternRecognition #MachineLearning https://arxiv.org/abs/1905.08233 http://youtu.be/p1b5aiTrGzY https://t.co/MKCoCcK7vX

0 replies, 6 likes


hatok: freaky

0 replies, 6 likes


codyb: this stuff just keeps getting crazier. feel like we are hitting some sort of flywheel where so many innovations are stacking on top of one another. via @casualdiego https://twitter.com/dmitryulyanovml/status/1131155659305705472?s=12

0 replies, 6 likes


⚕《Kira Powers》⚕: Greeeeat, now every dystopian Sci fi novel is possible, where people are framed for doing/saying things they never did. This could NEVER go wrong, right??? What a world...what a world...

1 replies, 6 likes


Mitch Rapp: 😱

2 replies, 6 likes


İdil: Creepy but cool!

0 replies, 6 likes


Film Learnin: This is some Ghostbusters 2 sh*t!

3 replies, 6 likes


ً: idk if this is a good idea

0 replies, 6 likes


Montreal.AI: "Few-Shot Adversarial Learning of Realistic Neural Talking Head Models" Zakharov et al.: https://arxiv.org/abs/1905.08233 #ArtificialIntelligence #DeepLearning #MachineLearning https://t.co/nEl2UhbD4A

0 replies, 6 likes


Tyler Saumur: This is actually so wild. Welcome to 2019 #AI

1 replies, 6 likes


The Muppet Jared Takes Duckburg: This is fucking insane. All hail Skynet.

2 replies, 6 likes


Joe Casabona 🎙: Soon we won't be able to trust any video as a source.

2 replies, 6 likes


Danny Jones: I don’t think people are worried enough about this. This will give plausible deniability for any real or fake video as this tech advances.

0 replies, 6 likes


Jason_Toff.exe: Things are about to get interesting...

0 replies, 6 likes


Gary M. Sarli♿️🎓🎲: .@brooklynmarie your job is going to suck so much worse now These are still in the uncanny valley (THE DEAD EYES), but it's just a matter of time.

1 replies, 6 likes


Parsa Zereshki: Pretty Crazy. by using #GANs to #animate a head based on as little head picture !

1 replies, 6 likes


Benjamin Haddad: This is scary

0 replies, 6 likes


Christian Mio Loclair: Wow crazy results. Congratz!!!

1 replies, 6 likes


Steve Christey Coley: tonight, in "We're All Gonna Die"

1 replies, 5 likes


Lexi 🛡️: You know, I always hoped most engineering courses taught you to focus on the real world implications of your actions. Oh well

0 replies, 5 likes


Black Hannity 🇺🇸: I wonder how long the CIA have had access to this tech and what they've done with it 🤔

0 replies, 5 likes


White✝️Stone: Reality has just become surreal. @babylonsingular ... thoughts?

1 replies, 5 likes


sara 🎶: imagine how much easier catfishing could be

0 replies, 5 likes


Dijital Habitat: ▶️ http://youtu.be/p1b5aiTrGzY 📝 http://arxiv.org/abs/1905.08233

0 replies, 5 likes


badidea 💫: Here’s the associated paper btw https://arxiv.org/abs/1905.08233

1 replies, 5 likes


🇮🇪 Tanners 🇻🇦: Wow can't see this going badly at all

0 replies, 5 likes


Sergii: Interesting paper, with pretty appealing demos - Few-Shot Adversarial Learning of Realistic Neural Talking Head Models - https://arxiv.org/abs/1905.08233 and more demos -https://www.youtube.com/watch?v=p1b5aiTrGzY #DeepLearning https://t.co/WCREywkFhY

0 replies, 5 likes


Ananda: Wow.

0 replies, 5 likes


Rafael Mauna 📀: Beware of the super easy to make fake videos... specially for your parents/older folks that will be targeted with this kind of crap. Not even mention the kids that could be baited with this...

0 replies, 5 likes


rixanne: aaaaaand we’re fucked

0 replies, 5 likes


Patriarch Prime | ☦️ ﷺ: Definitely isn't going to collapse the legal system at all

1 replies, 5 likes


Bindu Reddy: The revolution is just beginning... Using Transfer learning and generative modeling, Samsung managed to create talking heads from just one photo of a person. Now you can take a photo of person and make a video of them saying anything - https://arxiv.org/abs/1905.08233

2 replies, 5 likes


Eric: This is crazy! 😮😱 Animated talking heads from stagnant sources like art and pictures - via Samsung AI lab. Thanks @vgman94! #AI

2 replies, 5 likes


Jeff Winchell: Generating talking head videos from a single image was done a year ago. I wonder what will be possible in another year. "Few-Shot Adversarial Learning of Realistic Neural Talking Head Models" https://arxiv.org/abs/1905.08233 Maybe we can replace the talking heads on cable 🤣 https://t.co/EHHvHaTG0S

1 replies, 5 likes


literallyjustq: Deepfakes are going to be used in propaganda very very soon. We need a technological literate population to resist upcoming fake news. I mean if people could get convinced by #Momo, a fake only based on hearsay with a few still images, imagine how bad this will get?

1 replies, 5 likes


Ian Durham: This is creepy. Very creepy. And scary when you stop to think about the implications.

1 replies, 5 likes


Daisuke Okanohara: Few-show talking head model; extracting an embedding vector from few images and then apply it to the generator by AdaIN and train it with GAN conditioned with face landmarks. a pre-trained model can fine-tune well with few images. https://arxiv.org/abs/1905.08233 https://youtu.be/p1b5aiTrGzY?t=298

0 replies, 5 likes


🅼🅰🆁🅺 🅲🅾🅾🅺 = Mark Williams-Cook: 😳

1 replies, 5 likes


Lance A. Scott: Amazing...and creepy...technology from Samsung creates video from a single image. #AI http://arxiv.org/pdf/1905.08233… https://t.co/DuM6Y5u7vw

2 replies, 5 likes


Kevin Richard: wtf

0 replies, 5 likes


Andrew Barker: Just curious: what are the non-horrifying possible uses of this technology, exactly?

1 replies, 5 likes


Chine Ezeks 🧡⚡️: It's interesting watching humans invent stuff, knowing they will eventually abuse it & cause more unhappiness with it.

2 replies, 5 likes


troy_s: Just when you thought you’d never read the words “great paper” and “Samsung” in the same sentence...

0 replies, 5 likes


Béatrice LHUILLIER: Flippant 😮

0 replies, 5 likes


Stefan Larsson: Fascinating development. Now imagine the potential malicious uses along these lines. @aimllu @anamariadutceac @katjadevries

2 replies, 5 likes


Katya Sedova: This paper shows how ML can produce fake video from a few images. No large dataset needed. Before real-time detection/debunking catches up, the potential for exploitation and exposure to false information is vast. So is the threat of undermined trust in video as proof of fact.

0 replies, 4 likes


ara wilson: Arriving at Uncanny Valley. Please watch the closing doors.

0 replies, 4 likes


Zally: I'm scared D:

1 replies, 4 likes


James Gammell: our ability to determine what is real is going to be shattered at some point right

1 replies, 4 likes


Cullen Taniguchi, MD: This is super cool and also downright scary👇🏼😳

0 replies, 4 likes


Luiz Bento: O.o

0 replies, 4 likes


Michael P Gibson: Ok Dostoyevsky talking is scary

1 replies, 4 likes


Daniel 송-Marshall: Have you ever stopped and asked yourself, not if you could, but if you should?

1 replies, 4 likes


Marcus Troy: Hey guys stop using those filters... the world about to get scary!

0 replies, 4 likes


Omer F. Inanoz: Vay

0 replies, 4 likes


CyberRäuber: "Oh brave new world, that has such people in't!"

0 replies, 4 likes


Sylvain Baillet: @Samsung AI lab in #Moscow are pushing the limits of Adversarial Learning from one or few source images to animate talking heads, VERY, may almost TOO realistically. Please do not put this tech in all hands... #AIforGood Paper: https://arxiv.org/abs/1905.08233

1 replies, 4 likes


MʀWᴏᴏᴅᴏ: *Severe Dr Ian Malcolm face*

3 replies, 4 likes


barry: Telling me they haven't fooled us with this before?

0 replies, 4 likes


Luiza Caires: Wow

0 replies, 4 likes


Ian Quigley: Neat! Also, am sure it will never be misused.

2 replies, 4 likes


Rob Annan: Ok, no more trusting anything or anyone. This should be fun.

2 replies, 4 likes


HUMMUS HQ✨: this is fucking terrifying and definitely not a good thing

8 replies, 4 likes


ashkan soltani: Soon, we'll be able to re-animate our loved ones just from a photo.....

0 replies, 4 likes


Aaron Disney: Fake news has a wonderful new tool here #PutThatInYourPipe

0 replies, 4 likes


Shirley Zhao: A #preprint on @arxiv is top of the #altmetrictop100 list this year. Go #openscience #openaccess! Topic is super interesting as well... Few-Shot Adversarial Learning of Realistic Neural Talking Head Models https://arxiv.org/abs/1905.08233 https://t.co/uP0oGckbsa

1 replies, 4 likes


🦆🦆in a row: So don't believe frazzledrip when you see it?

2 replies, 4 likes


federella: I can picture far too many scenarios in which this is used to harass women. You know...the same way it happened with deep fakes.

0 replies, 4 likes


PEOPLE'S REPUBLIC OF REGGAETON 🇭🇳: There is gonna be a coup d’etat using this within 10 years.

0 replies, 4 likes


Oded Rechavi 🦉: The good news is that if there’s an embarrassing video of you doing something you can deny it’s really you.

0 replies, 4 likes


Clades Im: This shit LITERALLY makes video evidence worthless.

0 replies, 4 likes


chlo👑: this is a black mirror episode waiting to happen

0 replies, 4 likes


Chris Kelsey: Yeah.....this won’t be abused at all

1 replies, 4 likes


Micah Grimes: How long until courts have to start weighing the validity of video evidence because of this technology? Are they already? https://twitter.com/DmitryUlyanovML/status/1131155659305705472

1 replies, 4 likes


Genesis Node: How does the world determine video authenticity when this kind of AI and software exists?! I’m not an advocate of blockchain for everything but there might just be a case for a DLT based verification for media and news statements to stop “fake news” becoming the only news.

0 replies, 4 likes


Brenna 🔜 Congress 2019 at UBC: NO ONE ASKED FOR THIS

0 replies, 4 likes


Mitsuo Yoshida; AI bot: This tweet probably has the most retweets and favorites of the tweets that link to arXiv. https://twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 4 likes


Daddy Warpig: If you could imagine the future of the human race, it is #FakeNews Forever.

0 replies, 4 likes


Iswar Nethi: Imagine 5/10 years down the road, when the algorithms behind deepfakes have been properly constructed, an app or commercially available software could just scan a picture & let you watch videos of someone you've lost. My god.

0 replies, 3 likes


Alec Mazo: When will the deceased family members talk to us with the help of just a picture and some cognitive science AI?

2 replies, 3 likes


Digital Maverick: Good Lord. This is astonishing and truly terrifying

1 replies, 3 likes


Abuamerican: There's no way this can lead to much good

1 replies, 3 likes


Kung-hsiang, Huang (Steeve): @Samsung AI made Mona Lisa animate realistically with their newly published GAN model. Paper: https://arxiv.org/abs/1905.08233 #DeepLearning #MachineLearning #ArtificiallIntelligence #NeuralNetworks #GAN #SamsungAI https://t.co/5eq53nW23i

0 replies, 3 likes


Angry Beur ಠᴗಠ: Absolutely amazing! See how a realistic human head images can be obtained by training #convolutional neural networks to generate them... From one image! https://arxiv.org/abs/1905.08233 #AI #ArtificialIntelligence #machinelearning #neuralnetwork #NeuralNetworks #facepalm #Robotics https://t.co/k4fC8R99Ky

0 replies, 3 likes


Shohreh: The "many faced god" in act

0 replies, 3 likes


Bonnie Clark: It's becoming more important for gov to be able to demonstrate 1) a document is an authentic and trustworthy record of what it claims to be and 2) it has integrity and is complete and unaltered in all essential respects and free from tampering or corruption. #GCDigital #DigiPres

0 replies, 3 likes


Ch. Julian Villabona-Arenas: Amazing and terrifying at the same time

0 replies, 3 likes


Peter Welinder: I predict a magical future!

0 replies, 3 likes


Adrian: Should I be scared or fascinated?

2 replies, 3 likes


Unobomber™️: Essas coisas aqui são um exemplo de enciclopédia de “your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should”

0 replies, 3 likes


Enrico Gentili: We just scratched the surface of what GAN can do! And I’m not talking about animating a painting!

0 replies, 3 likes


Federico Cirett: Scary @Rql_Torres @lojesa @Talya @JorgeCirett @JAParedesCortes @hilarioolea @oleafontes

0 replies, 3 likes


Dilara Durak: Pretty cool to see Mona Lisa now has a living portrait. #deeplearning #AI YouTube: http://youtube.com/watch?v=p1b5ai… Paper: http://arxiv.org/abs/1905.08233 https://t.co/tE93CkNehj

0 replies, 3 likes


🌺 Caitlyn Meeks 🦋: Remarkable

0 replies, 3 likes


Dezi Esseintes: According to Captain Beefheart, there are only twenty-five people in the world and five of them are hamburgers.

0 replies, 3 likes


(((Greg Siskind))): Cool and scary. Seems like only a matter of time before it pops up in election interfering.

1 replies, 3 likes


Alex Rainert 🤦🏻‍♂️: Buckle up

1 replies, 3 likes


BioDecoded: Mona Lisa frown: Machine learning brings old paintings and photos to life | TechCrunch https://techcrunch.com/2019/05/22/mona-lisa-frown-machine-learning-brings-old-paintings-and-photos-to-life/ https://arxiv.org/abs/1905.08233 #MachineLearning https://t.co/1GADEZbwbj

0 replies, 3 likes


©️@ b o o s e 🇻🇦: The Japanese were right about cameras stealing your souls

0 replies, 3 likes


Anonymous Otto The Invisible: Slander is about to get interesting, huh?

1 replies, 3 likes


Jeremy Bratetich: What if as a species collectively we decided not to pursue this? Like thinking about it critically for 2 minutes and my mind has already said how about fuck this

1 replies, 3 likes


Triangle Filmmaking Community: Holy crap!

0 replies, 3 likes


Markus Sitzmann: Fascinating and scary at the same time.

2 replies, 3 likes


𝙅𝙤𝙨𝙚 𝙋𝙖𝙜𝙡𝙞𝙚𝙧𝙮: I worked on a special report about deepfakes earlier this year. This is a fascinating use case: making paintings come alive.

0 replies, 3 likes


Jaehaerys Khargaryen: Hooboy.

0 replies, 3 likes


PieDisliker: We got the talking paintings from Harry Potter now

0 replies, 3 likes


Zulfikar Ramzan: The bedrock of trust in digital content is already fractured. This research has the potential to annihilate that shaky foundation into oblivion. (Not to mention that these types of results can only improve over time...)

0 replies, 3 likes


Earl Operator: This shit is cool but mega dangerous, I don't support it

0 replies, 3 likes


Moritz: Photorealistic avatars from a single image 😲

0 replies, 3 likes


Philip Mai: Impressive. But eyes...they are still the window to the soul.

0 replies, 3 likes


Jason Wood: Nothing we see digitally will ever be trusted again

2 replies, 3 likes


Samantha Josephine Stockings: This is equally amazing and terrifying.

0 replies, 3 likes


Aubrey Bear: This... seems like a bad idea

0 replies, 3 likes


World Ethical Data Forum: By training convolutional neural networks, highly realistic human head images can be obtained. Training gen. requires large image-sets of a single person but after metalearning on a large dataset this system creates talking head models of prev. unseen people with a single image!

0 replies, 3 likes


reza mahmoudi: این مقاله ی که کانال تلگرامی (که قبلا معرفی کردم ) معرفی کرده عالیه Few-Shot Adversarial Learning of Realistic Neural Talking Head Models Zakharov et al.: https://arxiv.org/abs/1905.08233 #ComputerVision #GenerativeAdversarialNetworks #MachineLearning https://t.co/QVYWf9Nzex

0 replies, 3 likes


Brittany of Darnassus: Amazing tech, very concerning implications.

0 replies, 3 likes


Adam Jennison: The truth will become even harder to spot very soon! @nealasher maybe we adopt the Prador version of determining the truth?

0 replies, 3 likes


Hugo Fitzpatrick: This kind of technology is a threat to society. The kinds of abuses that this could inflict are not worth the novelty.

0 replies, 3 likes


David Nunez: whoa. Are you kidding me??

0 replies, 3 likes


Stefan Graunke: Wow!

0 replies, 3 likes


Manish Tayal: I don't know whether to be amazed or terrified.. #DeepFake #PostTruth #WhatIsReal #Technology

1 replies, 3 likes


Richard Torruellas: I know we are excited about stuff like this but I don’t think we fully understand the consequences.

2 replies, 3 likes


Tangier: @The_Petrichory @TurntableKittah @PollySpin I saw this today. https://twitter.com/dmitryulyanovml/status/1131155659305705472?s=21

1 replies, 3 likes


Vilson Vieira: That's amazingly disturbing! From https://arxiv.org/abs/1905.08233 https://t.co/SJOwpjK679

0 replies, 3 likes


Anne-Louise Quinton Art Crusader & ideas merchant: Ok this is really creepy and brilliant at the same time

0 replies, 3 likes


CityBoySizzle 🇪🇹: Imagine how dangerous this is. IMAGINE the implications.

0 replies, 3 likes


Peggy K: This is amazing and a bit scary.

1 replies, 3 likes


Besim Ismaili: Really interesting paper from Samsung AI lab - using GANs to animate a head based on as little as one target photo. Video: https://lnkd.in/d4dQrhP Paper: http://arxiv.org/abs/1905.08233 #AI #DeepLearning #GAN https://lnkd.in/diHhBwJ

0 replies, 2 likes


DeplorableStorm: Um, this is fucking terrifying. Anything is possible. [THEY] will try. Eyes on.

1 replies, 2 likes


ira.: that some black mirror shit

0 replies, 2 likes


Elemenopy: Wow

0 replies, 2 likes


bass: we can finally see mona lisa saying ni

0 replies, 2 likes


Shingai🔥: 😳

0 replies, 2 likes


Gary Unicorn: Nope. Don’t like this one bit.

0 replies, 2 likes


Ludo du Clary: #AI

0 replies, 2 likes


Yury Jajitzky: Looks like a "19 crimes" bottle... https://twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 2 likes


Shirley Zhao: Let's take a look at this year's #altmetrictop100... https://twitter.com/altmetric/status/1206895705396264963

1 replies, 2 likes


FeBu_MIC: Creepy and impressive at the same time... #fakevideos from paintings or historic photos with #deeplearning

0 replies, 2 likes


Florian Nebout: Not far from the uncanny valley, and It will cause a lot of ethics issues but Wow ! This is impressive...!

0 replies, 2 likes


Gregory: This coupled with ai voice copying is going to be scary, you won’t be able to confirm that anything you see or hear is real

0 replies, 2 likes


Andy Bhattacharjee: Ummmm this is craaaaa

0 replies, 2 likes


Dr. Vukosi Marivate: More important than ever for actors and actresses to own their likeness.

0 replies, 2 likes


Here comes another good Nickrob tweet:: I’m sure this will be used in good faith.

1 replies, 2 likes


💧Connor McMahon: This is so exciting, yet very scary tech. Do yourself a favour and watch this! What does this mean in the age of fake news?

1 replies, 2 likes


Toily Kurbanov: Call it Dostoevskian AI: meta-learning of adversarial generative models can train highly realistic living portraits with only 1 photo. @egorzakharovdl via @DmitryUlyanovML https://arxiv.org/abs/1905.08233 #AI #Dostoevsky #TGIF #FridayFeeling https://t.co/cnkkKTG9BQ

0 replies, 2 likes


Kristina Morss: This is both fascinating and terrifying.

0 replies, 2 likes


Jowanza Joseph: 👀

0 replies, 2 likes


Dave Swersky: How long before people start using the existence of deep fakes to dispute genuine video evidence? 🤔

0 replies, 2 likes


Arachnofool: So, finally we can ask her what is she smiling about! ;P (... after all this sophisticated art and tech, if she says 'fart', I'll quit humanity)

1 replies, 2 likes


Karim Muya: Few-Shot Adversarial Learning of Realistic Neural Talking Head Models #DeepLearning #MachineLearning #ArtificialIntelligence #DataScience https://arxiv.org/pdf/1905.08233.pdf https://t.co/g4JA6tLAxh

0 replies, 2 likes


gerald pao: Samsung AI moving towards small training sets for deep learning #DeepLearning #ai #DataScience

0 replies, 2 likes


Bruno Dusausoy: Can someone please tell me what’s a real-life use case of deep fakes ? I can’t think of any, except fooling and manipulating people.

4 replies, 2 likes


Ines: This will end well. 🙄

1 replies, 2 likes


INDEED Innovation: Uuhh creepy and amazing as well...

1 replies, 2 likes


Cat Staggs 🏳️‍🌈: the line for lawsuits against Samsung AI lab starts right over there...

0 replies, 2 likes


SegaMasterTim: Amazing!

0 replies, 2 likes


Fire Tracker: Creepy

0 replies, 2 likes


シャンディヤちゃん🎨: Oh no

2 replies, 2 likes


Serge McGraw: Can I please have one of me nodding ...would be a great replacement for many of my Skype and Zoom meetings!

1 replies, 2 likes


Yann Ranchere: Another this is freaking me out tweet

0 replies, 2 likes


David Feeney: Soon online #FakeNews will be joined by online #Fakevideo #auspol #ausdef @ASPI_ICPC

0 replies, 2 likes


Andréia Bohner: Impressive! #AI https://twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 2 likes


Ivan Arrizabalaga: Just impressive!

1 replies, 2 likes


Snowbird 🌼: ‘What is truth?’

0 replies, 2 likes


sɐɹɐןouʇǝɹ: hello future talking classic paintings museum

1 replies, 2 likes


Emma Price: Oh sh*t. Nothing is real. There’ll be no culpability if people can deny they ever said something that was caught on tape. Burn it all down.

1 replies, 2 likes


Ankur Modi: Fantastic work by @egorzakharovdl and @Samsung AI Team on using generative adversarial network (GAN)s to create living shots from static images. Full video taht includes Einstein at https://www.youtube.com/watch?v=p1b5aiTrGzY&feature=youtu.be (paper at https://arxiv.org/abs/1905.08233) #artificialintelligence #DeepLearning https://t.co/4ZHCfpEbpf

1 replies, 2 likes


Mustapha Itani: This is terribly dystopian on so many levels.

0 replies, 2 likes


Yvan Gauthier: That's really impressive.

0 replies, 2 likes


The Real General Zod: This will be the excuse for what is coming from the Weiner laptop.

1 replies, 2 likes


कर्मा भूटीया: @evankirstel Imagine this. Now you need only 1 photograph. https://twitter.com/lanceascott/status/1132457115161903104?s=21

0 replies, 2 likes


russjackson: Definitely nothing to be worried about here in an era of fake news & the best extremist & corporate propaganda money can buy...

0 replies, 2 likes


Sergey Minaev: Few-Shot Adversarial Learning of Realistic Neural Talking Head Models https://arxiv.org/abs/1905.08233

0 replies, 2 likes


Mr XBob 🌱 🔮 🏥 🦁: We're doomed. This and DeepFakes will absolutely destroy any semblance of truth.

0 replies, 2 likes


FritzCast Podcast: ...this is what John Connor meant when he asked "We're not gonna make it, are we?"

0 replies, 2 likes


Lee ''Dat Thang'' Baker: this is fucking terrifying

0 replies, 2 likes


Jevaughn Brown 👑+++: We are doomed af. Don't know how people without a sense of gallows humor are going to survive.

1 replies, 2 likes


Bernard Steer: 😮

0 replies, 2 likes


Magilla Gorilla: This is frightening as fuck. We literally almost can’t trust anything.

0 replies, 2 likes


Aram Vartian ➡️ GenCon: Scary, but all new tech is scary. I just want to get my hands on it and start a Civil War AP.

0 replies, 2 likes


𝔐𝔞𝔤𝔫𝔢𝔱𝔰 🧲: this has the potential for some incredible misuse, but it is also pretty remarkable

0 replies, 2 likes


Subba Reddy: @spectatorindex I foresee some interesting applications in the movie industry 😀😀 Few-Shot Adversarial Learning of Realistic Neural Talking Head Models https://arxiv.org/abs/1905.08233 https://techcrunch.com/2019/05/22/mona-lisa-frown-machine-learning-brings-old-paintings-and-photos-to-life/

0 replies, 2 likes


Rim K., PhD.: You guys...

1 replies, 2 likes


José M. Chia: Wow.

0 replies, 2 likes


Vicky Kerr: This is terrifying so many reasons.

1 replies, 2 likes


Cristian Vogel: The landmark driven Mona Lisa at the end of this presentation

0 replies, 2 likes


Torsten Reil: This is seriously impressive.

0 replies, 2 likes


Jin Hyun Cheong: Holy cow

0 replies, 2 likes


Jeremy Gordon: 🔥🔥This is fine. 🔥😗☕️🔥😬🔥🔥

2 replies, 2 likes


The Rolistes Podcast: Quite amazing/scary the visual world we are entering 😐

0 replies, 2 likes


Ryan Butner: The barrier-to-entry in regards to volume of needed training data for deep fakes is falling quite rapidly. Some of these are now even being generated from a single image.

0 replies, 2 likes


Machine Learning: RT @Sarah_Mojarad: With only a few starter images, Samsung AI Center technology can produce subject matter experts for evening news outlets. Research from @egorzakharovdl and colleagues: https://arxiv.org/abs/1905.08233 #AI #MachineLearning #Technology #scicom… https://t.co/rKZ9GSSTF1

0 replies, 2 likes


Chris Serpico: amazing and terrifying

0 replies, 2 likes


verort: Astonishing, but also shocking...#3dmodeling

0 replies, 2 likes


Kevin Muldoon: This is scary

2 replies, 2 likes


✨Heidi Quicksilver✨: Woah

0 replies, 2 likes


Kidz 'R EViL 🏁: Artificial Intelligence is a gift and a curse.

3 replies, 2 likes


Benjamin Wiederkehr: Don‘t see anything that could go wrong with this. Not a thing.

1 replies, 2 likes


Dub: This is some scary crap! The future will be filled with fake stuff. A lot of people will abuse this for the wrong reasons and it’ll be a big problem then.

0 replies, 2 likes


comfortwriter 📱🤓📝: Well, well, well... This definitely won't be used for evil or disinformation 🙄 You can now create whole videos simulated from an individual photo. 🤯

0 replies, 2 likes


Ian McKelvey: This is insane.

3 replies, 2 likes


Muhammad Aamir: Paper by #Samsung AI to use GANs to animate talking sequences ... @arXiv_Daily https://arxiv.org/abs/1905.08233 #ArtificialIntelligence #MachineLearning

1 replies, 2 likes


Femi Longe: So now a video of you can be "forged" using just your pictures... Wahala!!!!

0 replies, 2 likes


Astro Yuki 🌎🚀: Watch the YouTube video thru the link in the post. This is amazing tech but in the very near future the comment “video or it didn’t happen” may not work because videos can be “photoshopped” from a still image. Crazy!

1 replies, 2 likes


Peter Riebeek: Frightening stuff...

0 replies, 2 likes


John Walter: These videos of people speaking created from a single photo using AI are wild. The line at which we’re going to be able to tell fact from fiction is only going to get blurrier. #medialiteracy

0 replies, 2 likes


Israel Crisanto: Wow! I don't know if I am more excited or more worried. 🤤🤤🤤

0 replies, 2 likes


TheSickest ⏳: Oh, and speaking of observation of reality... —The year is actually 12,879AD and we are all in a temporal vacuum history class being taught in real-time as an MMOG were we all get to experience the evolution of AI and the collapse of western civ. Next week: Roman Empire.

0 replies, 2 likes


Natasha Lampard: I find this disturbing. Esp having just read this https://www.theguardian.com/technology/2019/may/24/facebook-leaves-fake-nancy-pelosi-video-on-site. *Just because you can, doesn’t mean you should.*

1 replies, 2 likes


Daricia 🇰🇳: creepiest shit I've seen all day...good work tho

0 replies, 2 likes


Mario Pastorelli: Am I the only one scared by this? Many people celebrating it but the only thing I can think about is how scary this is.

4 replies, 2 likes


Marcelo Ferranti: what kind of black magic is this?

0 replies, 2 likes


Hamid Shafeeu: 😯😲

0 replies, 2 likes


VonVictor V. Rosenchild | Founder and CEO: @FastCompany A @SamsungUS #MachineLearning - #ML #Algorithm is bringing the features of the Mona Lisa to life. Research Paper on @arxiv_org 🖥️https://arxiv.org/abs/1905.08233 @FastCompany Article 🖥️https://www.fastcompany.com/90354561/harry-potter-moving-portraits-could-be-real-with-samsung-ai?partner=rss&utm_source=twitter.com&utm_medium=social&utm_campaign=rss+fastcompany&utm_content=rss @TechCrunch Article 🖥️https://techcrunch.com/2019/05/22/mona-lisa-frown-machine-learning-brings-old-paintings-and-photos-to-life/ @HubBucket @BioDecoded https://t.co/3FNoKXNi5a

0 replies, 2 likes


🦝: Man, the future is going to be one fd up place https://twitter.com/dmitryulyanovml/status/1131155659305705472

0 replies, 2 likes


Frank Orlando: Jesus this is incredible and terrifying. This will NOT be used for good. Just to spread more misinformation.

0 replies, 2 likes


Mike Butcher: It might be fun to get the Mona Lisa to talk, but what problems might arise? New forms of fake news?

0 replies, 2 likes


Ginny Caughey: ICYMI seeing is no longer believing with deep fakes tech!  https://twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 2 likes


Drew Royster: On the positive end it would be cool to see Tesla give lectures in his voice with his face.

0 replies, 2 likes


Mo Elleithee: This is completely terrifying. Deep Fake misinformation campaigns are about to get a whole lot more sophisticated.

0 replies, 2 likes


Matthew Smith: We are so screwed.

1 replies, 2 likes


Pierre Richemond: StyleGAN + the below.

0 replies, 2 likes


🍒καρβουνάκι🍒: 😯

0 replies, 1 likes


Paolo Missier: The wonders and dangers of Unreality

0 replies, 1 likes


Chris Cox: Sci-fi is now sci-reality

0 replies, 1 likes


mary kathryn syms: 👀

0 replies, 1 likes


Clay R. Fuller, Ph.D.: What the future of “fake news” looks like, right here: 👇🏼

0 replies, 1 likes


Ryo Ishimaru👌: !

0 replies, 1 likes


H2Mass: I'm terrified

0 replies, 1 likes


mdubs: The future is both terrifying and exhilarating

0 replies, 1 likes


Frank Edoho: 😳😳😳😳

0 replies, 1 likes


Christian Varela: Sending this to people I follow to warn of this...terrifying tech: @ericselectrons @HPluckrose @ConceptualJames @peterboghossian @Emmyjewel @ChrisRGun @JaclynGlenn @shoe0nhead @RaychelTania @Silli_Scientist @miiowskii @millzebub @MsMelChen @KindOfaTrollLOL @AnsgarTOdinson @grngal

1 replies, 1 likes


Joni Job: target person AI-generated 😱

0 replies, 1 likes


Emelia 👸🏻: This is absolutely amazing tech, but also, have you thought about how it'll be misused @egorzakharovdl ?

0 replies, 1 likes


Matt Mehan: Digital verification will now become a function of institutional trust and vouched origin. Seeing will no longer be believing. (This assumes, of course, that these images are what they claim to be! ;)

0 replies, 1 likes


Austin: As mentioned before, you can’t even believe video CCTV evidence nowadays.

0 replies, 1 likes


Mr. Composition: The future looks scary

0 replies, 1 likes


B̵̢̢̹̭̳̦̬͐͐l̴̡͔͎̮̝̂̾u̷̲͋̓͒̀͆̈́͂̓̕e̴̤̙̩̳̭̙̮̍͊̒̅͝Nine: I wasn’t aware AI had progressed to this level, but here we are.

0 replies, 1 likes


Andrew Maltba: Wow 😯 #AI #Samsung

0 replies, 1 likes


Tyler Ford: This is absolutley terrifying. The average person does little to know background research on topics and takes things at face value. We are entering an age of flawless fakes and dangerous propaganda #MediaLiteracy #AlwaysDoYourResearch

1 replies, 1 likes


Jenessa Elizabeth: Wowzers is all I can say....one day the line between real and faux wil be so blurred we will marry cartoons and live happily ever after

0 replies, 1 likes


Tom: Technology is amazing (and scary...) the power of this sort of tech has so many possibilities

0 replies, 1 likes


Celine Hernandez: Both impressive and worrying at the same time...

0 replies, 1 likes


Javier Ramos Mejia: Be afraid, be very very afraid...

0 replies, 1 likes


Should This Exist?: #ShouldThisExist? h/t @_stevejacksonuk

0 replies, 1 likes


neuro.social.self: Moscow: head photos are being animated at the Samsung AI lab by @egorzakharovdl et al | May 2019 | see 📺 ▶️ http://youtu.be/p1b5aiTrGzY and 📝 http://arxiv.org/abs/1905.08233 https://t.co/H9xkxgB51S

0 replies, 1 likes


1mm tall fennec: We’re one step closer to the reanimated heads from Futurama

0 replies, 1 likes


Zala Bricelj - 🇸🇮 😊 #edutech is my jam: As much as this can be exciting in the sense of tech advancements, but raises more questions in terms of all the fake and deep fakes that can actually give a rise of mis-interpretation, false interpretation and ethical consequences..... 🤔

0 replies, 1 likes


Tina Liao: Woooooh

0 replies, 1 likes


Americana Prime 🇺🇸: Get ready folks, there’s gonna be so much fake news you won’t have any idea what’s real. There’s gonna be a fake video any day now of Trump saying something hateful. https://twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 1 likes


Vara: This is ingenious and also scary at the same time.

0 replies, 1 likes


Arun C Murthy: Inexorable march... towards? https://twitter.com/dmitryulyanovml/status/1131155659305705472

1 replies, 1 likes


Ilona Kickbusch: Scary

0 replies, 1 likes


अविभेद्य युवक: This is so fascinating and terrifying! One photo and a video of you saying anything can be made.

0 replies, 1 likes


Philip Laureano 🇭🇲: It's amazing when you realize that this was generated by two competing AIs where one was trying to spot a fake and the other one was trying to 'outfake' the spotter, all in an almost infinite loop.

0 replies, 1 likes


Shiloh: stop that

0 replies, 1 likes


Peacemaker: Holy shit 😱😱

0 replies, 1 likes


Sunny Virk: 😯

1 replies, 1 likes


LJ: @risetoflyy Education so your followers are fully aware that the chances they are getting punk'd with a phony video on social media have extremely high odds, including fake news video https://bigthink.com/technology-innovation/new-ai-can-create-fake-videos-of-people-from-a-single-picture download & READ PAPER https://arxiv.org/abs/1905.08233 Goes without saying for still images

0 replies, 1 likes


DataScienceNigeria: Samsung Deepfake AI can fabricate a video clip of you from a single photo, as AI animates photos without 3D modeling. Cautious! AI can make you appear to do or say things they never did. Watch: https://youtu.be/p1b5aiTrGzY Paper: https://arxiv.org/pdf/1905.08233.pdf Kudos @egorzakharovdl etal https://t.co/fj1SLgp3Qs

0 replies, 1 likes


l̴o̴o̴p̴u̴l̴e̴a̴s̴a̴: Black magic! Make a picture alive and speak with AI

0 replies, 1 likes


Jason 𝑻𝒉𝒆 𝑮𝒆𝒓𝒎 𝑮𝒖𝒚 Tetro: Whoa...

0 replies, 1 likes


Bill Merrill: Impressive and fascinating research, but I am only a fan of it academically. There are at least as many dangerous applications of this as there are healthy ones.

1 replies, 1 likes


Sam Hall 🏴󠁧󠁢󠁷󠁬󠁳󠁿🐾🐾🌈: Wow, so simple yet so powerful. AI in action.

0 replies, 1 likes


Sergeant 'stache: Our fake future will be glorious.

2 replies, 1 likes


IranContraSpacePirate: That is quite wild

0 replies, 1 likes


Sergio Quintana: Well this is mortifying. Be prepared for everything here to be faked. https://twitter.com/dmitryulyanovml/status/1131155659305705472?s=21

0 replies, 1 likes


The Original Lawdog: The attempt to discredit the upcoming incriminating videos has begun.

0 replies, 1 likes


Steven: It’s over

0 replies, 1 likes


Peter Samuelson: Interesting

0 replies, 1 likes


Geo the 🍀plorer 🚀🗼🗽: Unbelievable. Simply amazing tech. I'm gonna make my money talk and scold a few people 😎

0 replies, 1 likes


ppsc: where the fuck are you when we need you, Thanos?

0 replies, 1 likes


Michał Kilijanek: Am I the only one who is concerned?

0 replies, 1 likes


💚🕷Mika🕷💚: Holy fuck

0 replies, 1 likes


Saun: There are a finite number of linguistic sounds and corresponding facial expressions. Most of the English language can be mimicked with less than 30. Such was the basis of my masters thesis. It only took 20 years for technology to caught up to the research.

0 replies, 1 likes


🐯 Michael Trooder: This is huge. It puts whole new slant on quite a few things. It can make fake accounts harder to spot that we may have thought. And it also means that video interviews to establish identity might not be as definitive as we would like. This is what they Show us

1 replies, 1 likes


John Thicc: Privacy and human control over identity are pretty much over

0 replies, 1 likes


lynch 𓆏: Do not take part in Face ID blackmail

0 replies, 1 likes


DB: Yes, this definitely doesn’t seem to have potentially negative consequences 😳

0 replies, 1 likes


Vibujithan: GAN is sick. Few-Shot adversarial learning brings paintings to life. Paper: https://arxiv.org/abs/1905.08233 https://t.co/5zuamGmoPL

0 replies, 1 likes


Josh Stovall: Kinda crazy. If you think “fake news” is bad....Now we won’t even be able to trust videos we see if people talking.

0 replies, 1 likes


Evarwyn: Well this has terrifying applications in the wrong hands #JustSayin

0 replies, 1 likes


M G Harris 🧜(⧖): Wow.

0 replies, 1 likes


Vincenzo Tremonte: Holy crap.

0 replies, 1 likes


Brandon Christensen: Whoa. Like whoa.

0 replies, 1 likes


Emulation is not innovation: Now that is impressive stuff.

0 replies, 1 likes


Chris Kurdziel 🎮🤘🏻: If you are building a co around/in this space, I would love to chat!

0 replies, 1 likes


Socioemotional Selectivity Theory: AI is going to ruin us

0 replies, 1 likes


Fredösphere: Oooookay, I'd say the time to crank up that ol' Butlerian Jihad has pretty much arrived, wouldn't you agree guys?

0 replies, 1 likes


Jr.: Technology is getting scary.

0 replies, 1 likes


Christopher Ackerman: convolutional neural networks - Fake talking heads https://arxiv.org/pdf/1905.08233.pdf https://www.zdnet.com/article/samsung-uses-ai-to-transform-photos-into-talking-head-videos/#ftag=RSSbaffb68 https://t.co/ppjaN19xZg

0 replies, 1 likes


Daniel Susser: I’m starting to think technological determinists were onto something

0 replies, 1 likes


Temp: @joerogan Joe look at this https://twitter.com/DmitryUlyanovML/status/1131155659305705472?s=20

0 replies, 1 likes


Thomasorus @ Stunfest: This is both amazing and frightening.

0 replies, 1 likes


Mai ✨: This is INSANE

0 replies, 1 likes


RAF: Imma need to see everything IRL from now.

0 replies, 1 likes


Rob Jones: Oh no

0 replies, 1 likes


Ian Lewis: I'm kind of curious as to why research groups are pursuing this kind of research? Besides the downsides of making people look like they are saying things they never did, what benefits does this research provide?

0 replies, 1 likes


Cole of Course: Well this'll certainly save the police the effort of having to coerce those confessions outta people

0 replies, 1 likes


Gaurav Bengeri: Whenever there is a new disruptive technology, the world fears it at first and then slowly accepts it. However, I cannot see a world where this stops being scary. #ArtificialIntelligence #AugmentedReality #Biometrics #sciencetwitter #scicomm #

0 replies, 1 likes


Rob Thompson: this tech thing coming at us, it’s like a tsunami on the horizon. Trump, brexit, cambridge analytica etc was like the sea going out before the waves come in. Its about to overwhelm everything we know.

1 replies, 1 likes


Yassine BAKHTI: Woow, that's very dangerous 😲 👏

0 replies, 1 likes


ESFERARED: #FakeNews new dimension

0 replies, 1 likes


Jersey Gurl⭐️⭐️⭐️: The Deep Shyt; mean FAKE A.I. Technology; is improving?🤔🤡👁

1 replies, 1 likes


Wessel van Rensburg: Can you imagine how this could be misused in a country like South Africa?

0 replies, 1 likes


David Betz: Sure, 1984 and A Brave New World are required reading, but y'all need to add Black Mirror to your list.

1 replies, 1 likes


Marcus Bösch: Oh

0 replies, 1 likes


Gerhard Völkl: A real interesting #MachineLearnig application. You can create a movie from every picture. For good or bad intention.

0 replies, 1 likes


eric lab rat: Can we use this technology to have someone with self confidence make pitches for me on skype? My first pitch is for a modern update of Cyrano de Bergerac where a shy, ineloquent man uses AI technology to have a confident, well-spoken friend impersonate him to.......

0 replies, 1 likes


Fatima Rodriguez: Trying to get ahead of the narrative👇 Do you see now why CNN is screaming and throwing out "Doctored Video" all over the place. Were about to see some sick sruff on video and it will be real and it will be from everyone you trusted and admired.

0 replies, 1 likes


Tipsy Tai: Science fiction can no longer keep up with science fact

1 replies, 1 likes


Jose Hilario Hdez: AI surprises me again!! In a couple of years we'll have to attach certificates to every single video to proof authenticity. 😂

0 replies, 1 likes


Bob St. Jean: Let’s hope this tech isn’t abused... Oh, who am I kidding? We can count on it.

0 replies, 1 likes


Luanna: I'm ready for the talking portraits from Harry Potter to become real.

0 replies, 1 likes


Julianne Zimmerman: So incredibly cool. And terrifying.

0 replies, 1 likes


Berrak @ Berlin 🏳️‍🌈👩🏻‍💻☕️: yay more tools for racial inequality and sexual harassment

0 replies, 1 likes


The Salty Nerd: We are so screwed. Get ready for realistic threats from world governments using fake speeches and fake videos to induce panic. Question everything.

2 replies, 1 likes


Jen😏: This will have a bad ending.

0 replies, 1 likes


Lawliet: Top Three Worst/Most Immoral Inventions of Humans: 1) Deepfakes 2) DataGrid AI Models 3) This (Few-Shot Adversarial Learning of Realistic Neural Talking Head Models) 👇 AI was a mistake.

0 replies, 1 likes


Upandaway: Another step towards AI assisted generation of fake videos.

0 replies, 1 likes


Evert Bopp: Take a single picture and animate it to say anything they want. Looks like we've opened Pandora's of online lies and deception. #fakenews

0 replies, 1 likes


Nicolas: @LnaDtvk @MouynatA @mementoo @ECatalanist https://mobile.twitter.com/DmitryUlyanovML/status/1131155659305705472

0 replies, 1 likes


Paul RAD 🤞: Crazy and a bit frightening ! #IA cc @dr_l_alexandre

0 replies, 1 likes


lee ( li ): #AI #ThursdayThoughts Yikes!

0 replies, 1 likes


クッパ姫: first one looks like tina fey in drag

0 replies, 1 likes


Aaron Ng 🐉: Harry Potter shit right here

0 replies, 1 likes


akira: https://arxiv.org/abs/1905.08233 A paper on Few-shot learning at GAN combined with meta-learning. The facial expression is transferred from the Landmarks information of the face and the feature vector of the person extracted from Embedder. Mona Lisa talks in the demonstration. https://t.co/PvqWuMwW25

0 replies, 1 likes


james: this is terrifying

0 replies, 1 likes


Ramon: We will soon have no real use for the word "fake", referring to news or otherwise, as it will have become a distinction without a noticeable difference.

0 replies, 1 likes


Alex Empire 💋: Well this is fascinating and terrifying.

0 replies, 1 likes


MyLifeIsMunitz 🌹: Technology can be scary. How do we prevent a future where something’s reality may be close to unknowable by human comprehension?

0 replies, 1 likes


Rotimi Opeke: Wow. An absolutely stunning output from the Samsung AI Lab below. Random thoughts:

1 replies, 1 likes


Francois Orsini: Amazing...

0 replies, 1 likes


Angie Adamson: Oh no. Just when you thought someone filming the action might save you, along comes Samsung AI and blows it all out the water. Seeing as social media is the main news outlet for most people now, this could be rorted extensively. #auspol #AI https://twitter.com/DmitryUlyanovML/status/1131155659305705472

2 replies, 1 likes


Austin Benn: The definition of scary cool

0 replies, 1 likes


⚡️ ŁUKASZ WIECZOREK: *blink*

0 replies, 1 likes


Aslaug Gisladottir: This is really fascinating and scary at the same time

0 replies, 1 likes


❌REDACTED❌: Remember this. You’ll hear it again... a lot. All the stops are being pulled out. It lets you know how big this all is, also how dangerous it really is. DS fighting for its very existence. This would be a good story for a Matrix prequel. Real life isn’t real anymore.

1 replies, 1 likes


David Colquhoun: Ghastly potential for fake news merchants

0 replies, 1 likes


Galo Soler Illia: #miedo

0 replies, 1 likes


CAROVAN: I think we need to reboot Max Headroom for a comment.

0 replies, 1 likes


Lex Mercatoria: Scary stuff ...

1 replies, 1 likes


Dr Galloway: And here we are.

0 replies, 1 likes


Alex Helberg: Feels weird saying this, but I feel like this technology & others like it should legit be outlawed. I'm just having trouble imagining a single good thing coming from its use.

0 replies, 1 likes


Douglas Gomes: I hope we are prepared to deal with the amount of fakery that the future will bring.

0 replies, 1 likes


Lior Shkiller: Wow..

0 replies, 1 likes


Damián Fossi: This is fucking awesome

0 replies, 1 likes


Arturo Galván: Bye bye video clips as proof of anything

0 replies, 1 likes


Christopher Hoover: This is incredible

1 replies, 1 likes


Abdulla97: The future is a scary thing...

0 replies, 1 likes


Aki: We’re one step closer to reaching Hogwarts

0 replies, 1 likes


CHRISTOPHER HORTON strawman: @derekgilbert @FaceLikeTheSun @IlluminatiEyes @Obi1unome @Bin_Hamin @bWoStevie @StopTechnocracy

0 replies, 1 likes


Flat Lander 🇵🇱🐸👌🏻🥛📕: We will not need to erase our political opponents. They will erase themselves through the words we put in their mouths.

0 replies, 1 likes


Dondi West: Professor @daniellecitron, this is scary!

0 replies, 1 likes


Low Sock Ching: I don't know Japanese but if this is what I think it is it's mad impressive.

0 replies, 1 likes


Nariman Saftarli: 😐

0 replies, 1 likes


Ryan Horrigan: Feels like conversational, intelligent avatars embodied via *real-time* GANs will be possible. #GANs community, how far are we from real-time GANs (video)? We’re already (nearly) there with speech synthesis.

0 replies, 1 likes


Ahmad Nassri: this is how the Harry Potter portraits come to life ...

0 replies, 1 likes


Renee Hart: This type of technology needs to be illegal....think they can frame anyone of any crime

0 replies, 1 likes


Markus Trapp: Several recent works have shown how highly realistic human head images can be obtained by training convolutional neural networks to generate them. Few-Shot Adversarial Learning of Realistic Neural Talking Head Models: https://arxiv.org/abs/1905.08233 #ComputerVision #PatternRecognition https://t.co/fMOXUUWvVj

0 replies, 1 likes


johndczerwinski: Freaky use of #AI by animation using only a few photos! #ArtificialIntelligence #MachineLearning

0 replies, 1 likes


Úrsido Zurdo: 🤔

0 replies, 1 likes


Joe Sumner: Better stop believing it when we see it! Gulp.

0 replies, 1 likes


🧡: Shit like this is what keeps me up at night bro

0 replies, 1 likes


Ronald Plasterk: Animate heads using only few shots per person (or even 1). https://m.youtube.com/watch? Kijk in slot van deze video naar pratende Mona Lisa: feature=http://youtu.be&v=p1b5aiTrGzY#menu ▶️ https://youtu.be/p1b5aiTrGzY 📝 https://arxiv.org/abs/1905.08233 https://t.co/iy6Ki3LOzh

0 replies, 1 likes


David Davisson: #OurCreepyFuture

0 replies, 1 likes


rosa ★: man FUCK that

0 replies, 1 likes


Dan Stone: Cue Genesis, Land of Confusion.

0 replies, 1 likes


Nong Artrith: Interesting! Look at this: Samsung AI (Moscow) brought famous portraits to lives, more to read here: https://www.adweek.com/digital/samsungs-ai-lab-can-create-fake-video-footage-from-a-single-headshot. and https://arxiv.org/pdf/1905.08233.pdf.

0 replies, 1 likes


⚡️Max. Hwy.⚡️: No one mentioning Phillip K. Dick’s A Scanner Darkly scramble suits? Wow, tweeterverse you unread pile of technophiles 😅

0 replies, 1 likes


Dennis McNulty: This feels like the end of something.

0 replies, 1 likes


Trade War Wynne: Holy moley

0 replies, 1 likes


Bitt12: Overwhelming negative response here, but I think it's pretty cool :D

0 replies, 1 likes


AynsRedDog: Scary times people!! https://twitter.com/dmitryulyanovml/status/1131155659305705472

0 replies, 1 likes


Adrian Sanabria: Equal parts amazing, wonderful and terrifying! To be able to do this from an oil painting that's hundreds of years old... I'm dumbstruck.

0 replies, 1 likes


Jay Anaya: This is a great achievement in #ai more details https://youtu.be/p1b5aiTrGzY

0 replies, 1 likes


Lyle Coleman: @chadcestes @OctopusCaveman Someone should run these wax statues through that Samsung AI thing that brings pictures to life. I have a feeling it might be a wonder to behold. https://twitter.com/DmitryUlyanovML/status/1131155659305705472?s=19

0 replies, 1 likes


Tony Mac Lopez: This can not end well

0 replies, 1 likes


Gavin Purcell: holy shit. you though the whitney houston hologram was bad, wait for the jesus one.

1 replies, 0 likes


Carrie Schroeder: It's here.

1 replies, 0 likes


Manthos Kallios: This is TERRIFYING! We will no longer be able to trust anything other than first-hand experience.

1 replies, 0 likes


Jed Bracy: Wow - in light of today’s fake videos of Pelosi, and the inevitable rise of deep fakes, we’re really in for a new age of misinformation on steroids.

1 replies, 0 likes


𑀧𑀺𑀦𑀸𑀓𑀻: Interesting!!

1 replies, 0 likes


Karl Abi-Aad: @seriousssam interesting paper

1 replies, 0 likes


Nick Brown: Wow! This is how it starts. The End...

1 replies, 0 likes


Shoot Your Shot: They are not for us they are against us. They are not for us they are against us. They are not for us they are against us. They are not for us they are against us. They are not for us they are against us. They are not for us they are against us. THEY ARE NOT FOR US

1 replies, 0 likes


Kev: Framing people for crime is gonna way easier soon, especially in our beloved country.

2 replies, 0 likes


Stefan: Scientists have successfully made anime real.

2 replies, 0 likes


Richard Martin: I have mixed feelings about this. I can see it used for developing interactive narratives around art history - the subject talking to the viewer! But there's something really creepy and uncanny about it that I can't reconcile with somehow - maybe a Mirror of Erised sort of thing?

1 replies, 0 likes


Edward Fagan: Really impressive results from machine learning development, link to the paper here: https://arxiv.org/abs/1905.08233

1 replies, 0 likes


Gary: Well this is bloody terrifying!!

3 replies, 0 likes


Mike ⧖: So, yeah, this is terrifying.

1 replies, 0 likes


Zach Martin: I saw this already in Blade Runner 2049

1 replies, 0 likes


1 Raptor Installation(SN4): 😌😌 https://t.co/r7ntVAIhlI

1 replies, 0 likes


James Sweet III: nopenopenope

1 replies, 0 likes


Jessica Litwin: Uhm....

1 replies, 0 likes


Content

Found on May 27 2019 at https://arxiv.org/pdf/1905.08233.pdf

PDF content of a computer science paper: Few-Shot Adversarial Learning of Realistic Neural Talking Head Models