<h2>Touch Up</h2>
Thanks to a new neural network, convincing photo edits that would have previously required extensive Photoshop expertise has become as easy as drawing a quick doodle.
The AI algorithm, <a href="https://www.youtube.com/watch?v=dvzlvHNxdfI">recently shared</a> by YouTube channel &#8220;Two Minute Papers,&#8221; lets people quickly generate convincing, fake pictures of people just by sketching new features onto them and letting the AI handle the rest.
<iframe width="500" height="281" src="https://www.youtube.com/embed/dvzlvHNxdfI?feature=oembed" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
<h2>Smart Brush</h2>
The research, first published to the <a href="https://arxiv.org/abs/1902.06838">preprint server ArXiv</a> in February, is the product of computer scientists from South Korea&#8217;s Electronics and Telecommunications Research Institute. The paper includes examples in which the researchers tweaked photos of celebrities to have totally new expressions or facial features.
For instance, the researchers doodled a smile onto a photo of a neutral expression, and the algorithm churned out a photo that shows the subject smiling. Similarly, a couple brushstrokes enabled the neural net to edit people&#8217;s jawlines and hairstyles, and even erase sunglasses off their faces, generating realistic eyes in their place.
<h2>Maybe Don’t</h2>
While algorithms like this can be a fun way to tweak photos, and photo-altering algorithms could lead to better special effects in movies and video games, research like this can readily be misused to <a href="https://futurism.com/the-byte/lawmakers-deepfakes-undermine-public-trust">create misleading or harmful deepfakes</a>.
While manipulative media is nothing new, computer scientists seem hell-bent on developing the most convincing, easiest-to-use algorithms that <a href="https://futurism.com/ai-can-now-manipulate-peoples-movements-in-fake-videos">make pictures or videos</a> of people in situations that never really happened — and this new neural net is no exception.
WATCH MORE: <a href="https://www.youtube.com/watch?v=dvzlvHNxdfI">This AI Learned to “Photoshop” Human Faces</a> [YouTube]
More on deepfakes: <a href="https://futurism.com/darpa-68-million-technology-deepfakes">DARPA Spent $68 Million on Technology to Spot Deepfakes</a>

Painting a smile on top of creepy deepfake tech!

Amazing neural network turns doodles into realistic photos

Put A Smile On

A new AI algorithm can make creating deepfakes as easy as doodling a quick sketch on top of existing photos of celebrities.

Amazing Neural Network Turns Doodles Into Realistic Photos

Let's put a smile on that face!

<h2>Fake News</h2>
DARPA, the U.S. military’s research division, is hell-bent on fighting high-tech forgeries. In the past two years, it’s spent $68 million on digital forensics technology to flag them.
For DARPA, spotting and countering &#8220;deepfakes&#8221; — doctored images and videos that can be used to generate propaganda and deceptive media like <a href="https://futurism.com/ai-can-now-manipulate-peoples-movements-in-fake-videos">fictionalized political speeches or pornography</a> — is a <a href="https://www.cbc.ca/news/technology/fighting-fake-images-military-1.4905775">matter of national security, reports</a> the Canadian Broadcasting Corporation.
<h2>Losing Battle</h2>
For all of its hard work to spot edited videos, DARPA may still be fighting a losing battle. That&#8217;s according to Hany Farid, a computer scientist and digital forensics expert at Dartmouth College.
&#8220;The adversary will always win, you will always be able to create a compelling fake image, or video, but the ability to do that if we are successful on the forensics side is going to take more time, more effort, more skill and more risk,” Farid told the CBC.
The CBC acquired some videos and images of DARPA’s deepfake-spotting artificial intelligence in action and it’s not the most impressive highlight reel.
In one, the algorithm is able to go frame-by-frame to spot discrepancies. This is an impressive feat for an automated system, but the video in question is clearly edited — it’s nowhere near as sophisticated as the <a href="https://futurism.com/the-byte/lawmakers-deepfakes-undermine-public-trust">most advanced deepfakes</a> out there, which can imitate a <a href="https://futurism.com/the-byte/deepfakes-translation-tech-synthesia">real person’s voice</a> and facial expressions.
<iframe frameborder="0" height="404" src="//www.cbc.ca/i/caffeine/syndicate/?mediaId=1372526659778" width="718"></iframe>
<h2>Putting a Finger on it</h2>
Other examples obtained by CBC showed off DARPA’s algorithm spotting inconsistencies in lighting and other subtle cues that give away a doctored video.
People who want to create misleading deepfakes may always be a step ahead of the people trying to stop them, but DARPA’s digital forensics program still has another two years of research ahead of it.
As long as they stay on top of new deepfake techniques and keep their algorithms on the cutting edge, they may be able to spot all but the most advanced, high-budget and high-tech productions.
That way, when a video of Donald Trump announcing a nuclear strike plunges the world into the apocalypse, we’ll at least know it was the real thing.
READ MORE: <a href="https://www.cbc.ca/news/technology/fighting-fake-images-military-1.4905775">A new &#8216;arms race&#8217;: How the U.S. military is spending millions to fight fake images</a> [CBC]
More on DARPA’s deepfake research: <a href="https://futurism.com/darpa-stop-deepfakes-facebook-google">If DARPA Wants To Stop Deepfakes, They Should Talk To Facebook And Google</a>

But the deceptive videos will always be one step ahead.

DARPA spent $68 million on technology to spot deepfakes

Seems Legit

DARPA spent $68 million over two years on deepfake-spotting AI. It's making progress, but the government might be fighting a losing battle.

DARPA Spent $68 Million on Technology to Spot Deepfakes

New military algorithms can tell whether a video was doctored, but DARPA think it’s losing the fight.

<h2>Loaded Language</h2>
On Thursday, a U.K.-based tech firm released a remarkable demo video of its flagship software.
&#8220;I&#8217;d like to demonstrate how Synthesia technology can make me speak 7,100 languages fluently,&#8221; says the smiling person in the clip. A few seconds later, a digitally manipulated version of the clip shows the same person saying the same sentence in French, Mandarin, and Portuguese.
But while the clip seems lighthearted enough, the technology powering it could be anything but — in fact, some see it as a direct <a href="https://www.theguardian.com/technology/2018/jul/27/fake-news-inquiry-data-misuse-deomcracy-at-risk-mps-conclude" target="_blank" rel="noopener">threat to democracy</a>.
<iframe width="500" height="281" src="https://www.youtube.com/embed/PhZpnDqNNmY?feature=oembed" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
<h2>Your Words, Not Mine</h2>
According to <a href="https://www.synthesia.io/" target="_blank" rel="noopener">Synthesia&#8217;s website</a>, the purpose of its AI-powered dubbing tech is to create a world without language barriers in which anyone can enjoy any video content, regardless of its language of origin.
&#8220;We hope that this new medium will foster cultural exchange, joy and deeper understanding in the same manner that the written word has done for centuries,&#8221; the company writes on its homepage.
Others are far less idealistic. They envision a future in which malicious actors use software like Synthesia&#8217;s to create deepfakes — videos in which a person appears to say something they didn&#8217;t actually say.
“By blurring the line between fact and fiction, deep fake technology could undermine public trust in recorded images and videos as objective depictions of reality,” three members of Congress <a href="https://futurism.com/the-byte/lawmakers-deepfakes-undermine-public-trust" target="_blank" rel="noopener">wrote</a> in <a href="https://schiff.house.gov/imo/media/doc/2018-09%20ODNI%20Deep%20Fakes%20letter.pdf">a letter</a> to Daniel Coats, the U.S. director of national intelligence, in September.
<h2>The Deep End</h2>
Examples of deepfakes are already online, but they&#8217;re typically either <a href="https://www.wired.com/story/these-new-tricks-can-outsmart-deepfake-videosfor-now/" target="_blank" rel="noopener">not that realistic-looking</a> or created by <a href="https://youtu.be/cQ54GDm1eL0" target="_blank" rel="noopener">trained digital artists</a>. Synthesia&#8217;s AI-powered software could give anyone with a computer the ability to create their own convincing deepfakes.
While Synthesia has acknowledged this possibility, the company doesn&#8217;t appear too troubled by it.  “We are trying to solve a very complex and technical problem,” chief executive Victor Riparbelli <a href="https://www.telegraph.co.uk/technology/2018/10/27/tech-threatens-democracy-funding-uk-taxpayers/" target="_blank" rel="noopener">told the Telegraph</a>. “We are not releasing any software to the public…There is a wider discussion to be had about the malevolent use of this kind of stuff.”
Of course, not actively releasing software to the public and actually keeping it out of the public&#8217;s hands are two very different things — and if this particular software gets into the wrong hands, there&#8217;s no telling what kind of devastation it could cause.
READ MORE: <a href="https://www.telegraph.co.uk/technology/2018/10/27/tech-threatens-democracy-funding-uk-taxpayers/" target="_blank" rel="noopener">Tech That &#8216;Threatens Democracy&#8217; Is Being Funded by UK Taxpayers</a> [The Telegraph]
More on deepfakes: <a href="https://futurism.com/the-byte/lawmakers-deepfakes-undermine-public-trust" target="_blank" rel="noopener">Lawmakers: Deepfakes Could &#8220;Undermine Public Trust&#8221; In &#8220;Objective Depictions of Reality&#8221;</a>

This software could let you speak 7,100 language. It could also rip apart society.

See the translation tech lawmakers say “threatens democracy” in action

Speak No Evil

In a new video, tech company Synthesia demonstrates its AI-powered dubbing tech, which could make it easy for others to create deepfakes.

See the Translation Tech Lawmakers Say “Threatens Democracy” in Action

It could make it easy for anyone to create deepfakes.

<h2>DON’T PROGRAM WHEN YOU’RE HUNGRY</h2>
Artificial intelligence and human intelligence aren&#8217;t the same, to be sure. But it seems we have one big thing in common: we spend our time fantasizing about dinner.
In <a href="https://arxiv.org/pdf/1809.11096.pdf">a research paper</a> published Friday in the preprint server arXiv, a team of AI researchers from Google DeepMind teamed up with a scientist from Heriot-Watt University to develop what they’re calling the largest, most advanced Generative Adversarial Network (GAN) ever. And to prove that it works better than any other, they had it create photorealistic images of landscapes, (very) good dogs and other animals, and of course some hot, juicy burgers.
<h2>WHAT’S THE BEEF</h2>
<a href="https://futurism.com/artificial-intelligence-video-games">Generative Adversarial Networks</a> are one of the more sophisticated types of AI algorithms out there. In short, one network creates something — in this case an image — as realistically as possible. Meanwhile, another network checks its work against examples of the real deal. This back and forth makes the two networks gradually improve to the point where the latter system is really good at detecting AI-generated images but the other is still able to fool it.
GANs are generally used to create media, whether its a new level of a video game or 3D models. And though their ability to fool us and themselves presents a bit of a double-edged sword, their ability to discriminate algorithmic from human output can be used to find and fight misleading deepfakes.
This process is what allowed DeepMind’s burger-cooking GAN to go from creating, <a href="https://qz.com/1408576/artificial-image-generation-is-getting-good-enough-to-make-you-hungry/">as Quartz reported</a>, a weird, beefy blob in 2016 to what actually looks like an appetizing (albeit overcooked) slab of burg today.
<h2>ALTERNATIVE PROTEIN</h2>
Some argue that we need to move away from red meat, but AI isn’t ready to join us yet. While the burger-generating algorithm is in great shape, others aren’t quite there. For instance, the twists and turns of brass tubing that make up a French horn baffled the network. The butterfly image is just a little off, and any attempt to render a photo of a human results in a horrifying blob monster.
READ MORE: <a href="https://qz.com/1408576/artificial-image-generation-is-getting-good-enough-to-make-you-hungry/">In just two years, AI has managed to make a slightly more appetizing cheeseburger</a> [Quartz]
More on GANs: <a href="https://futurism.com/darpa-stop-deepfakes-facebook-google">If DARPA Wants To Stop Deepfakes, They Should Talk To Facebook And Google</a>

Google used its most sophisticated AI yet to create pictures of burgers.

Dinner Time

Google DeepMind built the largest-scale generative adversarial network to date. The algorithm can make photorealistic pictures of burgers and dogs.

Google Used Its Most Sophisticated AI yet to Create Pictures of Burgers

The singularity will be hot and juicy.

<h2>FAKE NEWS</h2>
In the early, optimistic days of the internet, we thought it would be a repository of high-quality information. Instead, it&#8217;s starting to feel like a bottomless ocean of lies that rewards attention-grabbing disinformation and pollutes the political process.
That&#8217;s the note of alarm that three members of Congress sounded in <a href="https://schiff.house.gov/imo/media/doc/2018-09%20ODNI%20Deep%20Fakes%20letter.pdf">a letter</a> this week to Daniel Coats, the U.S. director of national intelligence. In it, the lawmakers warned specifically about the technology called deepfake, which lets computer users with little tech savvy create convincing footage of people doing and saying things that they never actually did.
&#8220;Hyper-realistic digital forgeries — popularly referred to as &#8216;deep fakes&#8217; [sic] — use sophisticated machine learning techniques to produce convincing depictions of individuals doing or saying things they never did, without their consent or knowledge,&#8221; read the letter. &#8220;By blurring the line between fact and fiction, deep fake technology could undermine public trust in recorded images and videos as objective depictions of reality.&#8221;
<h2>MISINFO</h2>
Representatives Adam Schiff (D-CA), Stephanie Murphy (D-FL) and Carlos Curbelo (R-FL) signed the letter to Coats. In it, they requested that the heads of the intelligence community prepare a report that would tell Congress what steps it has planned to fight the dissemination of faked clips.
&#8220;Forged videos, images or audio could be used to target individuals for blackmail or for other nefarious purposes,&#8221; they wrote. &#8220;Of greater concern for national security, they could also be used by foreign or domestic actors to spread misinformation.&#8221;
<h2>INFOCALYPSE</h2>
Deepfakes rose to prominence early this year on Reddit, where posters started using it to splice the likenesses of <a href="https://motherboard.vice.com/en_us/article/bjye8a/reddit-fake-porn-app-daisy-ridley">celebrities into pornographic films</a> and the visage of Nicolas Cage into <a href="https://theoutline.com/post/3121/you-can-now-watch-nicolas-cage-play-james-bond?zd=1&amp;zi=wnrybjaa">movies he never appeared in</a>. Soon afterward, experts at a DARPA meeting of media forensics experts became concerned about it the technology. One expert <a href="https://theoutline.com/post/3179/deepfake-videos-are-freaking-experts-out?zd=3&amp;zi=xxbsop7s">told the Outline</a> that doctored footage of a world leader declaring war could spark a &#8220;full-blown nuclear holocaust.&#8221;
A deepfake hoax of a world leader hasn&#8217;t viral — yet. If it does, it will be a test of our collective skepticism — in an age when even genuine information is swiftly politicized online.
READ MORE: <a href="https://schiff.house.gov/imo/media/doc/2018-09%20ODNI%20Deep%20Fakes%20letter.pdf">Deep Fakes Letter</a> [Adam Schiff]
More on information warfare: <a href="https://futurism.com/darpa-stop-deepfakes-facebook-google/">If DARPA Wants To Stop Deepfakes, They Should Talk To Facebook And Google</a>

The internet is starting to feel like a bottomless ocean of lies that rewards attention-grabbing disinformation and pollutes the political process.

Lawmakers: deepfakes could "undermine public trust" in "objective depictions of reality"

Artificial Intelligence

Lawmakers warned this week about "deepfakes," a technology can create convincing footage of people doing and saying things that never happened. 

Congress: Deepfakes Could Undermine "Objective Depictions of Reality"

Welcome to the infocalypse.

Lawmakers: Deepfakes Could “Undermine Public Trust” in “Objective Depictions of Reality”

There are already fake videos on the internet, manipulated to make it look like people <a href="https://www.youtube.com/watch?v=cQ54GDm1eL0">said things</a> (or <a href="https://motherboard.vice.com/en_us/article/gydydm/gal-gadot-fake-ai-porn">appeared in porn</a>) that they never did. And now they&#8217;re about to get way better, thanks to some new tools powered by artificial intelligence.
Instead of just moving a source video’s lips and face, an artificial intelligence-powered system can create photorealistic videos in which people can sway, turn their heads, blink their eyes, and emote. Basically, everything that an actor does and says in an input video will be translated into the video being altered.
<iframe width="500" height="281" src="https://www.youtube.com/embed/qc5P2bvfl44?feature=oembed" frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe>
According to the <a href="https://arxiv.org/pdf/1805.11714.pdf">research</a>, which will be presented at the <a href="https://s2018.siggraph.org/">VR filmmaking conference SIGGRAPH</a> in August, the team ran a number of tests comparing its new algorithm to existing means of manipulating lifelike videos and images, many of which have been at least partially <a href="https://futurism.com/darpa-stop-deepfakes-facebook-google/">developed by Facebook and Google</a>. Their system outperformed all the others, and participants in an experiment struggled to determine whether or not the resulting videos were real.
The researchers, who received some funding from Google, hope that their work will be used to improve virtual reality technology. And because the AI system only needs to train on a few minutes of source video to work, the team feels that its new tools will help make high-end video editing software more accessible.
The researchers also know their work might, uh, worry some folks.
“I’m aware of the ethical implications of those reenactment projects,&#8221; researcher Justus Thies <a href="https://www.theregister.co.uk/2018/06/04/ai_neural_network_fake_faces/">told The Register</a>. &#8220;That is also a reason why we published our results. I think it is important that the people get to know the possibilities of manipulation techniques.”
But at what point do we get tired of people “raising awareness” by further developing the problem? In the paper itself, there is just one sentence dedicated to ethical concerns — the researchers suggest that someone ought to look into better watermarking technologies or other ways to spot fake videos.
Not them, though. They’re too busy making it easier than ever to create flawless manipulated videos.

The AI system can create the best manipulated video yet.

AI can now manipulate people's movements in fake videos

A new AI algorithm can create the most advanced deepfakes yet, complete with emotions and gestures, after just a few minutes of training.

/Deepfakes

Put A Smile On

Amazing Neural Network Turns Doodles Into Realistic Photos

Seems Legit

DARPA Spent $68 Million on Technology to Spot Deepfakes

Speak No Evil

See the Translation Tech Lawmakers Say “Threatens Democracy” in Action

Dinner Time

Google Used Its Most Sophisticated AI yet to Create Pictures of Burgers

Artificial Intelligence

Lawmakers: Deepfakes Could “Undermine Public Trust” in “Objective Depictions of Reality”

Artificial Intelligence

AI Can Now Manipulate People’s Movements In Fake Videos

Load More