"And governments realise this could be used as a weapon." Weaponising deepfakes "A lot of governments still don't know what deepfakes are and I'm here to explain what it is, which things are possible," he says.
Since then, Ume has been talking to governments too, advising them on the rapid evolution of deepfake capability. To quell some of the hysteria, Ume went on a media blitz, revealing how he made the videos and reassuring the world his intentions were only to entertain. "Because we never exposed ourselves in the beginning, people didn't know who created these videos," Ume says. "You had a lot of articles talking about the end of the world … and, 'This technology is getting out of hand.'" Then Ume and Fisher released their creations into the wilds of the internet. The reaction was completely unexpected, as online viewers debated whether the real Tom Cruise had in fact joined TikTok. It still took over two months to build the computer model, then dozens more hours ironing out the glitches frame by frame using digital graphics software. Ume collected over 6,000 images of the Hollywood actor taken from different angles, with different facial expressions, to train the algorithm. "For a deepfake, you start with the source data - I mean pictures and videos of the character you want to deepfake.
#Mr deep fake how to#
"I did a few months of research, how to do it, and a half-year later I had my first deepfake," he says. Where others were exploring the darker side of deepfakes, Ume saw the creative potential. Ume first got interested in deepfakes in late 2018 when he saw a news report about how the technology was being used for malicious ends. Their partnership would soon go on to produce some of the most convincing deepfakes ever made. "People were amazed, like, 'How did he do this?'" Ume recalls.
The resulting video, uploaded to social media, got the reaction the pair were after. Fisher wanted Tom's "concession" speech to use deepfake technology for the big reveal, where Tom Cruise would seem to emerge dripping wet from a swimming pool, laughing manically and draped in an American flag. When Fisher started planning the sequel, he turned to Ume for help. "It was a funny video where Tom Cruise was running for president," Ume says. In response to Professors Chesney and Citron's blog, Herb Lin, a senior cyber policy research scholar at Stanford University, suggested technology vendors could create "digital signatures" that are assigned to the purchaser.Īlthough, as Dr Lin pointed out, encrypted keys already inserted into digital cameras have been cracked.But the man in the video is not Tom Cruise, it's Miles Fisher, an actor who created a spoof campaign clip called "Run Tom Run" in the lead-up to the 2020 US presidential election. How do we authenticate content?Ī watermark or a digital "key" to identify authentic content could be a useful tool, suggested Dr Harandi.īut there has been little movement towards a global protocol on these matters. He suggested America's Defense Advanced Research Projects Agency should create a "secure internet protocol" to authenticate images. This way people will know whether the voice or image is real or from an impersonator," Congressman Ro Khanna told The Hill. "We all will need some form of authenticating our identity through biometrics. "Public trust may be shaken, no matter how credible the government's rebuttal of the fake videos." Teeth are notoriously hard to synthesise realistically.īut the technology will improve, and quickly.īeyond the morality of porn "deep fakes", an altered video of Donald Trump, for example, could have serious geopolitical consequences.ĭoctored videos could show politicians "taking bribes, uttering racial epithets, or engaging in adultery", suggested American law professors Bobby Chesney and Danielle Citron on the Lawfare blog.Įven a low-quality fake, if deployed at a critical moment such as the eve of an election, could have an impact. If you pay attention, you can see lip movements don't entirely match the speech. German researchers have controlled Vladimir Putin's face.ĭr Mehrtash Harandi, a senior scientist who researches machine learning at Data61, said the output of these deep-learning machines is often blurry. A public crisisīarack Obama has been made to lip sync. And so has the search for new ways to sort fact from fiction. The democratisation of machine learning has begun. "It was 100 per cent predictable," said Hany Farid, a computer science professor at Dartmouth College, who specialises in digital forensics.ĭr Farid thinks we are at a crossroads - we can still tell when videos have been doctored. Researchers use "Real-time Face Capture" on Russian President Vladimir Putin.
0 kommentar(er)
