Tom Cruise Deepfake Videos Can Be a Security Threat: TikTok Tom Dubbed a 'Terror' [Report]

A viral TikTok video of Tom Cruise reignites fears for Deepfake videos. The video was surprisingly realistic that even experts are becoming afraid of it.

Deepfake brings out the amazing potential in videography. However, its system could easily be weaponized by malicious actors.

Several deepfake videos of Cruise are going viral on TikTok. One of it features the Hollywood superstar playing at a golf course. The overall video, including the actor's facial expression and features, is too realistic that it convinced most of the internet users that Cruise himself had joined the TikTok community. In a matter of days, the video spiked from 2.5 million views to over 10 million, making "TikTok Tom" a global phenomenon.

The graphics artist Chris Ume, and his collaborator Miles Fisher, also felt surprised by the total of views.

Tom Cruise Deepfake Videos

This was not the first deepfake Tom Cruise video the duo posted. However, even professionals have a hard time differentiating this current TikTok video from the real man himself.

Deepfakes are Artificial Intelligence-generated videos that utilize a person's facial structure to recreate expressions. The more accurately it reads on a person's face, the more realistic the generated video gets. ABC Net reported that the technology first emerged five years ago, featuring low-quality impersonations of iconic figures like Obama.

However, in these last 18 months of continuous developments, Deepfake is getting capable of producing lifelike fakes. Ume, along with a few other handpicked deepfake wizards, also worked on a TV series by Matt Stone and Trey Parker. The series called "Sassy Justice" deepfakes Donald Trump behind a cravat and grey wig.

Deepfake Videos Can Be a Security Threat

Ume said that building these convincing deepfake videos still requires a lot of time, skill, and resources. For the "Tik Tok Tom" videos, he collected over 6,000 images of the Hollywood star. It also takes him two months to build the computer model and a dozen more hours to remove glitches. By estimate, it takes around 24 hours to create one minute of the deepfake video.

Nonetheless, Deepfakes is opening up serious threats for being weaponized. Former CIA officer and disinformation specialist Matt Ferraro was quoted by ABC Net as saying: "I think that 'terror' is probably not too strong a word... It's because they realize how dangerous [deepfakes] are. It does seem like it's really going to be a fundamental challenge to the information environment."

The ABC report also cited sourced in Washington calling it a "terror."

Deepfake could easily be exploited in social media and political parties. Malicious actors could deepfake videos of leaders like U.S. President Joe Biden declaring war. Regardless of how the government handles the situation, the video would still create mass confusion for the general public, potentially inciting conflict and casualties.

Unfortunately, Deepfake videos are hard to detect in most system algorithms. The Verge reported that Facebook, as the winning algorithm to spot challenging real-world examples, could only detect the video at a 65.18 percent accuracy. Fortunately, systems up to date are working hard to overcome this security risk.