Home / Business and Politics / Worrisome deepfake: Tesla, Pope Francis or Neymar could also be in your ad

Worrisome deepfake: Tesla, Pope Francis or Neymar could also be in your ad

Recently, there has been increasing talk about the problem of videos featuring celebrities or, more precisely, their deepfakes, which have begun to appear in various advertisements and campaigns, with or without their permission. Digital simulations of entrepreneur Elon Musk, actors Tom Cruise, Leonardo DiCaprio, and many other well-known faces have appeared in advertisements, mostly without their knowledge.

The machine learning company Paperspace Co. recently released a promotional video for its company featuring Cruise and DiCaprio, with neither of the two having spent a second filming that campaign, nor have they ever agreed to support the company in such a way. However, this did not prevent Paperspace Co from promoting itself with these world-famous faces. Of course, the promotional videos of that company were created using deepfake technology, which employs computer-generated representations to make Hollywood and business dignitaries say and do things they have never said or done.

Although many of these deepfake ads will not deceive a vigilant viewer or at least someone who understands technology and realizes that DiCaprio, Cruise, and similar figures will not give their likeness for free to just anyone, some ultimately remain deceived. Therefore, analysts believe that the increasing acceptance of deepfakes could profoundly reshape the media and marketing landscape, and it has also opened up new legal and ethical questions.

Authorized and Unauthorized

Of course, if someone agrees to have their likeness included in some marketing campaigns or promotional videos using deepfake technology, there are no problems – this practice is called authorized deepfake. Marketing experts say that authorized deepfake would allow celebrities and well-known figures to appear in campaigns without the need to physically show up on set or in front of cameras, which would reduce costs and open new creative possibilities for those running campaigns, while also saving actors some time, which means a lot to everyone. However, the problem arises with unauthorized deepfakes because, as things stand now, it falls into a gray legal area, so industry experts agree that celebrities should fight against the spread of unauthorized digital reproductions of themselves and manipulation of their brand and reputation.

American lawmakers began addressing the phenomenon of deepfakes in 2019 when the state of Virginia banned the use of deepfakes in so-called revenge pornography. Soon after, Texas banned them in political campaigns, and California banned deepfake technology in both cases. Last year, the U.S. National Defense Authorization Act mandated the Department of Homeland Security to produce annual reports on the threats posed by this technology. So while some direction towards regulation has been initiated, there is still no law addressing the use of deepfakes in advertising.

Low Probability of Lawsuit

Aaron Moss, head of the litigation department at the law firm Greenberg Glusker, says that celebrities have had some success suing advertisers for unauthorized use of their images under so-called publicity rights laws. However, deepfakes did not exist at that time, and he cites Woody Allen’s $5 million settlement with American Apparel in 2009 due to the director’s unauthorized appearance on a jumbo billboard advertising that clothing brand.

Recently, the startup reAlpha used Elon Musk in its deepfake video where Elon explains the concept of crowdfunding while in a bubbly bath. The first Musk video was released a few days after reAlpha launched its public fundraising offer, and the video ultimately garnered 1.2 million views on YouTube and attracted interest in reAlpha from ’22 thousand people in 83 countries,’ said Christie Currie, the marketing director at the mentioned company. She added that the company ‘avoided directly linking the video to its fundraising efforts.’

image

photo

– There is always a small risk with any type of parodic content, but generally, if it is intended to be educational, satirical, and if you have disclaimers in place, there should be no problems, as long as you do not force monetary transactions – emphasizes Currie.

It is a thin line. The likelihood that someone like Musk would sue a startup for a deepfake video is low, and small startups might decide that the risk is worth the significant publicity it would generate for them, believes attorney Moss. However, the ease of creating deepfakes means that some celebrities could soon find themselves in a pile of ads showcasing their unauthorized but very convincing likenesses, Moss adds. He concludes that any attempt by aggrieved parties to attack small startups or individuals creating deepfakes would be ‘death by a thousand cuts.’

Namely, the language in contracts written years before the technology existed can be vague enough to allow companies to use existing footage to create new deepfake videos. Therefore, actors, athletes, and other celebrities will begin inserting clauses that prohibit any new such use of their likenesses in all commercial contracts they sign. Companies currently most often seek deepfake videos of celebrities for internal use for training, communication, or entertainment, but not for advertisements, notes Daynen Biggs, owner of Slack Shack Films. He reveals that client requests can sometimes be quite bizarre.

– One client requested a video with former President Donald Trump in the role of Mr. Potter, the wealthy villain in the movie ‘It’s a Wonderful Life’ – notes Biggs, acknowledging that deepfake has the potential to be extremely harmful, but he says they always ensure that what they create is not harmful and focus on fun and entertaining ways to share messages.

However, experts and practitioners say that deepfake technology will become increasingly popular in advertising, as it can help brands and agencies produce more content faster while eliminating many production costs, which is certainly a great lure for using this technology for advertising purposes.

Artificial History

At the largest creator of AI-generated videos in the world – Hour One, based in Tel Aviv and New York, the same company that allows you to convert your text into video through its program and also get a virtual presenter/actor, they are very angry when someone calls deepfakes that way.

They say the term has taken on a derogatory meaning due to the frequent misuse of footage, vulgar and false montages, and they are among the honest ones, those who offer authorized deepfakes, i.e., authorized video content created using AI, but with footage of individuals or characters who have allowed such use. Their strategy leader Natalie Monbiot stated that they do not like the word deepfakes, saying that they have been authorized from the beginning and describes Hour One as a company that has ‘built a legal and ethical framework for collaborating with real people to create their likeness in digital form.’

image

photo

She, like her company, is against the unauthorized use of deepfakes and is aware of the damage that, let’s call them, illegal deepfakes can cause. And the damage is also discussed by Eric Horvitz, Microsoft’s chief scientist, who says that ‘deep fake threats’ are on the rise and that they include interactive deepfakes, which offer the illusion of conversing with a real person, and compositional deepfakes, with bad actors creating a larger number of deepfakes to assemble something that Horvitz calls ‘synthetic history.’ He believes that more fake videos could be incorporated to follow some narrative.

Imagine a fictional terrorist attack that never actually happened, but with a large number of deepfake recordings, many would think it did.

The Business Side of Deepfakes

A bright future is also predicted for the business side of using deepfakes. The research and consulting firm Forrester recently published its predictions for the field of artificial intelligence for 2023, one of which is that 10 percent of Fortune 500 companies will generate content using artificial intelligence tools. The report mentions startups like Hour One and Synthesia, which ‘apply artificial intelligence to accelerate video content generation.’

Another report predicts that in the next five to seven years, as much as 90 percent of digital media could be synthetically generated. The business side of the deepfake story is ‘very underrated,’ emphasizes Victor Riparbelli, CEO of the London-based company Synthesia, which creates video content using artificial intelligence. The company was founded in 2017, has over 15,000 clients, a team of 135 people, and is growing double digits every month.

Among its clients are fast-food giants, including McDonald’s, research company Teleperformance, and global advertising holding WPP. Synthesia states that its artificial intelligence is trained on real actors. It offers images and voices of actors as virtual characters that clients can choose to create educational videos on anything they need. Riparbelli notes that the fast-food chains they work with must train hundreds of thousands of people every year. From how to stay safe at work, how to handle customer complaints, how to operate a fryer, he explains, and says that such deepfake videos are an excellent way to adopt new knowledge.

image

photo

Previously, Riparbelli says, companies could shoot a few such videos that they could not later edit, and other employee training was likely done through PowerPoint slides or PDFs. Today, such an approach cannot work, he believes, especially with younger generations. This was also recognized by the international language school Berlitz, one of the first business clients of Hour One, which needed to digitally transform after 150 years of offering traditional classroom learning. Now Berlitz, after collaborating with Hour One, has over 20,000 videos in various languages created using Hour One’s deepfake technology.

Tagged: