MIT uses deepfake tech to show ‘Nixon’ giving moon disaster speech
Jul 22, 2020, 9:55 AM
51 years ago this week, humanity took one giant leap onto the lunar surface. Astronauts Neil Armstrong and Edwin “Buzz” Aldrin became the first humans in history to stand on the moon.
Those first successful steps into the cosmos represented the culmination of the combined genius, effort and ingenuity of teams here on earth.
Hope for the best, prepare for the worst
But what if those small steps hadn’t been successful? What if the Apollo 11 astronauts never made it back to our little blue planet?
It’s estimated that more than 600 million people watched the moon landing live in 1969. But NASA and President Richard Nixon had prepared for the possibility that Armstrong, Aldrin and their command module pilot, Michael Collins, wouldn’t make it back.
‘In event of moon disaster’ – Read the speech President Richard Nixon never gave
In the month before launch, Apollo 8 astronaut Frank Borman approached President Nixon’s chief of staff, William Safire, to suggest the government prepare for what to do if the worst happened.
As a result of that conversation, Safire drafted a speech titled ‘In event of moon disaster.’
Richard Nixon never gave that speech… until now.
What is a deepfake?
In its most simple form, a deepfake is an altered video. Using deep-learning artificial intelligence (AI) programs, a computer can be fed a source image (like an actor) and a target image (like a speech in the Oval Office by Richard Nixon) and combine the two.
“The word ‘deepfake’—a combination of ‘deep learning’ and ‘fake’—refers to any media that uses artificial intelligence to simulate people doing or saying things they never did or appearing in fabricated situations without their consent,” MIT’s Joshua Glick explained.
In Event of Moon Disaster poster (MIT Centre for Advanced Virtuality )
You may have already seen some of these videos. Many times they’re used to put a favorite actor somewhere they shouldn’t be, like casting Nick Offerman as every character in Full House. But security experts say that the same technology can be used to sow misinformation.
That’s not why MIT’s Center for Advanced Virtuality created this Nixon deepfake project though. They say they didn’t make the deepfake to add to conspiracy theories about the moon landing, but to empower the public.
“Experiencing a powerful use of new technologies in a transparent way has the potential to stay with viewers and make them more wary about what they see in the future. By using the most advanced techniques available and by insisting on creating a video using both synthetic visuals and synthetic audio (a ‘complete deepfake’), we aim to show where this technology is heading—and what some of the key consequences might be,” Panetta and Burgund said.
MIT’s deepfake Nixon Moon speech
“What you are about to see is not real.”
With iconic real images of the Apollo 11 mission in the background and the voice of CBS’s Walter Cronkite giving an update just before launch, the video produced by MIT and released this week mixes the real with the fake.
MIT used two different AI programs to computer-generate Richard Nixon inside the Oval Office, making his mouth move with words he never said.
“Usually we painstakingly code and construct our projects ourselves, as we are both sticklers for detail. But this time we were going to farm out the work to artificial intelligences (AIs). We could sit back, relax, and have the work done for us by a clever, highly trained black box,” Franchesca Panetta and Halsey Burgund, two of the directors of the project said.
“We were wrong.”
Between the two programs and the teams behind them, it took the group three months to make a six-minute video. Panetta and Burgund found that comforting.
“Making a high-quality deepfake still isn’t easy. However, researchers predict that manipulated images and videos that appear ‘perfectly real’ will be possible for everyday people to create within six months to a year.”
You can see more of how the MIT team created the Nixon deepfake video here.
