Source: Freepik

Deepfake

Deepfake, a combination of deep learning from Artificial Intelligence (AI) programs and fake, is a technique of synthesizing human images or sounds based on artificial intelligence techniques that use simulations or computer-generated replicas to create videos of people, celebrities and other personalities, saying and doing things they never said or did.

“The more advanced deepfakes use an autoencoder, which is a deep learning AI program tasked with studying various video clips to build a broader understanding of what that person would look like from various angles and in various environments. The source of the study is then mapped to the target within a video. 

Other methods use a different type of machine learning, known as Generative Adversarial Networks (GANs). A GAN is used to identify and enhance any flaws in deepfakes to reduce detection and make it harder to decode.” – Loop Digital – The Rise of Deepfake Technology

Legal and ethical issues of deepfake technology

The biggest threat of deepfakes is their ability to spread false information that appears to come from trusted sources, generating misinformation and political manipulation.

In an unauthorized way, deepfake technology can be used to achieve greater visibility through the unauthorized use of celebrity images or to create fake customer testimonials.

“The language of contracts written years before the technology existed may be vague enough to allow marketers to use existing images to create new deepfake videos.” – The Wall Street Journal– ‘Deepfakes’ of Celebrities Have Begun Appearing in Ads, With or Without Their Permission

There are two ways to create deepfake videos: through an original video source of the target, where the person is forced to say and do things they never did; or by changing the person’s face in a video of another individual.

Unfortunately, deepfakes are used to blackmail and create reputational damage for people, the most common use being non-consensual deepfake pornography (or revenge porn).

Another unfortunate and reprehensible use of deepfakes is when a target image is placed in an illegal, inappropriate, or compromising situation, such as lying to the public, engaging in explicit sexual acts, or using drugs.

Fake evidence involves the fabrication of fake images or audio that can be used as evidence of guilt or innocence in a legal case.

Deepfakes are also used as fraud, such as someone impersonating another person to obtain personally identifying information, such as bank accounts and credit card numbers. Or even impersonating company executives to access confidential information.

Forged deepfake materials are used to affect a company’s stock price. This is called stock manipulation. Imagine that a fake video of a CEO making damaging statements about your company is circulating on the internet. What might happen? It can probably reduce your stock price. The opposite effect, i.e. increasing a company’s stock, can occur in the face of the dissemination of a fake video about a technological breakthrough or product launch.

Also, the US Department of Homeland Security’s Growing Threat of Deepfake Identities report cited text messaging as a future use of deepfake technology.

How to detect deepfakes

According to the Tech Target website, there are some signs that the content may be deepfake, although AI is consistently outperforming some of those indicators, such as natural blinking.

On video deepfakes:

• Unusual or awkward facial positioning;
• Unnatural facial or body movement;
• Unnatural coloring;
• Videos that look strange when enlarged
• Inconsistent audio;
• People who don’t blink.

In textual deepfakes:

• Spelling errors;
• Sentences that do not flow naturally;
• Email addresses of suspicious origin;
• Phrases that do not correspond to the alleged sender;
• Messages out of context, irrelevant to any discussion, event, or problem.

Advantages of authorized deepfake technology

In art, deepfakes are commonly used to generate new music using the existing bodies of an artist’s work.

In events and also in the arts, deepfake technology can help recreate objects or people anywhere in the world at the same time.

In other industries, we see examples of the legitimate and legal use of deepfakes, such as in:

• Audio and entertainment in video games: with the manipulation of actors’ voices in certain scenes;
• Telephone customer support: with the use of fake voices for simple tasks, such as checking an account balance or registering a complaint; and
• Personalized user response applications: such as call forwarding and receptionist services.

Deepfake Advertising

The world of deepfake advertising is coming this decade – this is revealed by research from the New Swinburne University of Technology published in the Journal of Advertising: Preparing for an Age of Deepfakes and AI-Generated Ads: A Framework for Understanding Responses to Manipulated Advertising.

Source: Freepik

How deepfake ads are progressively gaining ground in the advertising world, becoming a phenomenon

Fact: AI-generated ads and manipulated advertising can produce deepfakes that are completely indistinguishable from reality.

Using authorized deepfakes can reduce costs for companies as it helps brands produce more content and with greater speed, by, for example, not requiring the person to be present on the production set at all times. In this case, the company buys a license to be able to use the identity of a celebrity and use digital recordings with that same celebrity to insert the desired dialogue, thus creating new videos.

Another advantage of using this technique is the possibility of hyper-personalizing the ads, creating models that exactly match the ethnicity, and physical characteristics, such as physical structure and skin color, type or style of clothing of the person watching those ads, etc. And why does this matter? Because consumers tend to buy more when they see products as extensions of themselves. This personalization using deepfake technology will help brands increase inclusion and reach wider audiences.

Source: Freepik

“These personalized ads can lead to more sales and improved brand reputation, as long as they don’t turn into customer hypervigilance – leading to privacy concerns and feelings of consumer vulnerability.” – Colin Campbell – Associate Professor and Researcher in digital innovations in marketing and advertising

Deep learning technology can also generate ads that use location data to localize ads, manipulating the content so that the person or celebrity speaks in their local language, references local attractions, etc.  

And what about omnichannel marketing campaigns? With deepfake technology, it will be possible to edit and retarget content much more easily at the expense of various media types.

On the other hand, deepfake ads can be intimidating to most consumers generating discomfort and more distrust and disbelief for their increasing use in advertising in general.

Especially if brands start inducing customers to buy by featuring fake UGC (User Generated Content) or influencer promotions as social proof.

The biggest concern for consumers is not being able to distinguish between what is real and what is fake.

Source: Freepik

“Deepfakes are likely to be widely used in mainstream media within a decade. (…) Marketers will strive to create “authentically human” deepfakes to look ‘real’.” –  Sean Sands – Professor and Researcher in retail and social media innovation

How deepfake ads use artificial intelligence and machine learning to generate convincing and extremely realistic replicas

This CNBC Television video, Number of deepfake ads increasing, published October 28, 2022, shows WilmerHale cybersecurity lawyer Matthew Ferraro talking about deepfakes, computer-generated images or videos of real people, which at the time were beginning to appear in advertising.

I must agree with Neil Patel, who said that deepfakes are getting a lot of bad publicity, but he believes that technology can, yes, bring good things to the world and that when you combine technology with marketing, with positive intent, we can forever change the way we communicate with our customers.

The malaria awareness ad in the Malaria Must Die campaign with David Beckham in 2019 is a great example of a dynamic campaign with influencers to increase reach, as deepfake technology allowed Beckham to speak in nine languages.

The Dalí Museum in St. Petersburg, Florida, USA, meanwhile, created a Dalí deepfake greeting visitors and interacting with them via several screens in an experimental video installation that used a type of deep learning technology.

Opportunities and concerns raised by the use of celebrity deepfakes in ads, movies and TV

In January 2023, the world’s first TV show using deepfake technology, the six-episode series Deep Fake Neighbor Wars, produced by Tiger Aspect in collaboration with synthetic media company StudioNeural, premiered in the UK.

In this article from LBB – Little Black Book, experts in intellectual property and technology law talk about the legal concerns over the use of deepfakes in advertising and entertainment, particularly about the series in question.

“This type of program tests the boundaries. It would need to be made very clear that the deepfake images are not real and also that the program is not sponsored or approved by the individuals portrayed.”– Ron Moscana – Partner at international law firm Dorsey & Whitney

Thus, it is clear that the principles of freedom of speech should be ensured, protecting the creator or producer of the content from liability, as long as the program clearly states that the people present there are not real or that real people have not endorsed it.

“A person can be held liable for using deepfake technology to infringe another entity’s intellectual property rights or a person’s rights of publicity or privacy. And the technology itself may be protected by intellectual property rights. Malicious use of deepfakes can also constitute fraud, defamation, identity theft, and other civil and criminal violations.” – Ryan Meyer – Of Counsel at the international law firm Dorsey & Whitney

Ron Moscana explains that even if there is no copyright for a person’s image, a celebrity can protect the right to commercially exploit his image (his name or visual ‘likeness’) only if he can prove that his image is recognizable and has some commercial value or that he is already exploiting it.

He goes on to say that, alternatively, “privacy and data protection laws may also be invoked to oppose unauthorized exploitation of a person’s name or photo.

The future of deepfake technology in advertising and entertainment

“There will likely be litigation over licensing disputes, particularly while the technology and these licensing agreements are new.” – Ryan Meyer – Of Counsel at international law firm Dorsey & Whitney

Ryan Meyer believes that as deepfakes become more common in advertising and entertainment, celebrity endorsement may become standardized in future entertainment contracts, including “twin digital licenses for promotional or merchandising purposes” alongside the main deal agreement.

 

How generative artificial intelligence will affect the business model of advertising agencies

According to this article from Branding in Asia, the use of generative AI in ad agencies ranges from content creation to enhanced personalization, analytics, and even improving the briefing process.

Chris Greenough, general manager of GrowthOps in Malaysia, details three types of agencies that may emerge in the era of generative AI.

1) Agency with “hand-made creatives and codes”

This type of agency will end up being a niche market, and will offer a level of personalized service to some clients who appreciate more authentic offerings that are “handmade” by humans and bring a special touch.

2) Infinite Agency

An agency founded by highly disruptive and innovative marketing experts with just a few employees, who will create large-scale advertising campaigns, digital products, and more with just their imagination and generative AI tools. Chris Greenough believes that Infinite Agency will also be a niche model.

3) Hybrid Agency

Hybrid Agency will be inevitable as, according to Greenough, agencies are working to incorporate processes that take advantage of a hybrid human technology approach that will improve efficiency and productivity. Greenough believes that agencies with a culture of strategic thinking are well positioned to succeed in this hybrid model, betting that this will be the dominant model as it offers services that combine human talent and creativity with the power of AI technology.

“The “war for talent” will shift from purely digital or technical skills to strategic ones. Agencies with an ingrained culture of thinking and strategy will succeed in using AI as a search tool. The ability to challenge ideas and stand out from the sameness will differentiate agencies.”- Chris Greenough –  general manager of GrowthOps in Malaysia

Deepfake Advertising: added value, if used ethically and responsibly

In general, deepfakes are legal, but there is a growing concern worldwide. In the US, deepfakes are only illegal if they violate existing laws, such as child pornography, defamation, or hate speech.

In advertising, the use of deepfake technology as a marketing tool, although it can create a culture of distrust or even generate annoyance in consumers, if used well, responsibly and ethically, and transparently by companies, can generate unique and engaging content that entertains, delights and even goes viral, including a more dynamic consumer journey.

“The way forward must include an ethical framework when it comes to technology design, development, and deployment. At every step, leaders have the option to elevate principles such as autonomy, fairness, respect, privacy, inclusion, and environmental stewardship.” – Beena Ammanath – Technology Leader, Trust, Ethics, and Global Head of the AI Institute at Deloitte

If used for good, it is undeniably a creative and innovative technology for marketers that multiplies their efforts, increasing brand loyalty and also sales.

There is no going back. Hyper-real experiences will continue to grow with a high potential to deliver more immersive marketing experiences.  

One thing is certain: regulatory decisions and technologies are needed quickly to keep things under control in this new era of deepfakes and artificial intelligence-generated ads.