If you’ve been scrolling through TikTok lately, you might have come across some random videos of Tom Cruise — but it’s not actually him.
The “Mission: Impossible” actor has been seen on the platform doing magic and showing off his golf skills. The videos circulating have garnered a whopping 11 million views in total. However, the action movie star hasn’t been on the app at all. It was revealed that this TikToker Tom Cruise was a product of deepfake technology. The technique uses artificial intelligence to swap one’s face on someone else’s body.
We are woefully unprepared for this.
AI safety needs to be bumped up our list of priorities.
— Mckay Wrigley (@mckaywrigley) February 25, 2021
Internet users were alarmed to see how highly-polished the impersonation was, even getting his voice down perfectly. The TikTok page @deeptomcruise recently went private after the videos started gaining attention from media outlets.
People mostly took issue with it because it poses as a concern for cyber security. If deepfake technology could convince us that it was Tom Cruise, imagine what it could do to political figures. It also puts regular citizens at risk of being incriminated for something they may not be guilty of.
I wouldn't be surprised if we see a lot more of that in 3 years.
— Malcolm Poindexter (@malcolm_p) February 26, 2021
Apparently some users have already been playing around with the technology. This TikTok shows a deepfake of Vladimir Putin, the long-time president of Russia.
wow
— sexytxt (@sexytext) February 26, 2021
Is it high time that technologies that are ripe for abuse be regulated already?
Other POP! stories you might like:
Vegan TikToker is banned for bigotry-fueled agenda against non-vegans
TikTokers believe that Texas snow is ‘fake’ and ‘a government plot’
From blackface to treadmill runways: A review of America’s Next Top Model
The auditory phenomenon called Backmasking, unmasked
‘Flat Tom’ rug design depicts iconic scene from the Tom and Jerry show
The "signatures" they detect are effectively the training data for the next iteration of faking AI.
Soon such fakes will be indistinguishable to all but a similarly trained and equivalently powered detecting AI and those who operate them.
— Kasilas (@kasilas) February 26, 2021
