News

Facebook, Microsoft, start, match, detect, deepfake videos

(Reuters) – Facebook Inc. is working with Microsoft Corp., to the nership at the AI-association and academics from different universities to start a match, in order to better detect the deepfakes, the company said in a blog post on Thursday.

FILE PHOTO: Attendees walk past a Facebook logo during the Facebook Inc. F8 developers conference in San Jose, California, united states, April 30, 2019. REUTERS/Stephen Lam

The social media giant is putting $10 million into the “Deepfake the Detection Challenge,” which aims to encourage the discovery of research. As part of the project, Facebook has been the commissioning of research to produce a realistic deepfakes to create a data set to be used for the testing of the apparatus.

The company said that in the video, which will be released in December, it will have paid for the actors and for which there is no information about the user can be used.

In the run-up to the US presidential election in November, 2020, the social media have been under pressure to address the threat posed by deepfakes, that is, to make use of artificial intelligence to create a hyper-realistic video in which a person seems to say or do something that they don’t.

While there is not yet a well-made deepfake video with huge political consequences in the United States, with the potential to be manipulated the video to be the cause of unrest, it was recently shown by a cheapfake clip from the House of representatives, Nancy Pelosi, manual, delayed and her speech seems to be unclear.

In August, the Democratic National Committee, showed that the threat of deepfake a video of her very own President, Tom Perez, to be in the audience at the hacker convention Def-Con-think that is the real Perez had skyped into the meeting.

“They deepfakes), lowering the bar for an adversary who wants to create a modified media,” said Matt Turek, who runs DARPA’s Media in-Store program.

Some of the researchers are working on systems for the authentication of a video or a picture, at the time of recording by means of digital watermarks. However, the rapid development of the deepfake of technology has led to an arms race between the deepfake-makers and those trying to detect the video.

“It’s a cat-and-mouse game. If I change the design of a detection system for deepfakes, I am giving the attacker a new receiver to test it against,” said Siddharth Garg, an assistant professor of computer engineering at the University of New York in the ‘ s, Tandon a School.

The technology is also becoming more and more accessible to less experienced developers. Last week, a Chinese app that is called His, which allows the user to have a convincing way of changing their views of the movie stars that rose to the top of the country in the app store, although it will also have to deal with the have an impact on the protection of privacy.

Some of the online deepfake makers are also tapping into this market to make it easy to deepfakes. The Machine-learning enthusiasts from countries from Poland to Japan, and to make it easier for people to get access to the custom deepfakes. They will have to upload a step-by-step YouTube tutorials, and charge $30 for 50 words or so of an AI-powered Home in an imitation of the voice, and it’s run by a self-service web sites that churn out deepfakes.

The Deepfake Detection, the Challenge is not the first time that Facebook, which currently does not have a specific policy with regard to the deepfake video, it has funded scientific research to the respondents.

House Intelligence Committee Chair Adam Schiff insisted in July that Facebook, Twitter, myspace Inc. and Alphabet, Inc., Google, which owns YouTube, to share their plans to address deepfakes. Facebook said that it will spend $7.5 million to the teams at the University of California, Berkeley, and the University of Maryland and Cornell University to respond to the threat.

One of these teams is carried out by the university of Berkeley, Professor Hany Farid, is the building a ‘soft biometric models, which are allocated to the politicians’ facial features Senator Bernie Sanders’ eyebrows jump to be a Senator, Elizabeth Warren, head spinning, to detect if a video is fake.

Facebook’s new contest, which will build on its links with these researchers, and the involvement of scientists from Cornell Tech, MIT, and the University of Oxford, UC Berkeley, and the University of Maryland, College Park and the University-at-Albany-SUNY.

In a statement on Thursday, the Ship, called the game “a promising step forward.”

Reporting Elizabeth, Illustrator; Editing by John Mitchell and Lisa Shumaker

Follow us

Don't be shy, get in touch. We love meeting interesting people and making new friends.

Most popular