And how marketers can battle deepfakes to protect their brands

The conversation surrounding Taylor Swift at the Super Bowl has completely overshadowed a battle she and millions of others are fighting. Within a single month, Taylor Swift has been the subject of deepfake videos twice, underscoring the risks faced by millions. It reflects the new world we live in, where we run immense risks of being deepfaked, our brands appropriated by AI impersonators and companies compromised. It's a stark reminder of the dangers presented by the AI Era.

Deepfakes represent an insidious threat that can damage public personalities in frightening ways. In fact, it has the potential to also damage our democracy, capitalism, brands, our personal lives and even our memories too. Don’t get me wrong, I love the Super Bowl Taylor Swift story but we mustn’t forget about her deepfake battles and what they mean. Learn about the two recent examples and discover what you can do to protect your brand, your celebrity partners and yourself.

Sexually Explicit Images of Taylor Swift

A few weeks ago, AI-generated, sexually explicit images of Taylor Swift spread rapidly across social media platforms, disturbing her fans and sparking a wave of protest and concern. One of the images on X, viewed 47 million times over a 17 hour period before the account's suspension, exemplified the issue, leading X to suspend several accounts posting the fake images, though they continued to circulate on other platforms. Swifties rallied on X, flooding it with protest posts to suppress the images. Reality Defender, a cybersecurity firm, confirmed with 90% confidence that these images were AI-generated using diffusion models, highlighting the ease with which deepfakes can be created and spread.

This incident underlines the growing concern about AI's role in generating nonconsensual pornography, a dark, evolving threat to privacy and dignity. Keep in mind that 96% of deepfake videos online are nonconsensual pornography. Despite some platforms' efforts to remove these images/videos and sanction the accounts responsible, the rapid spread and visibility of deepfakes have outpaced current measures to contain them, leading to recent calls for stronger legislative action against the creation and distribution of deepfakes. Yet, as AI technology advances, the challenge of curbing the production and dissemination of deepfakes becomes increasingly difficult, highlighting the need for a concerted response from technology companies, legislators, law enforcement officials, corporations and the public at large together.

Fake Taylor Swift Le Creuset Endorsement

And if that’s not all, earlier in January, Taylor Swift’s impersonation was used to collect consumer data and scam consumers. Taylor Swift's genuine appreciation for Le Creuset cookware, highlighted through her home décor and a Netflix documentary, contrasted sharply with the fake ads featuring her unauthorized endorsement of the brand's products on Facebook. These ads, part of a broader trend of celebrity-focused scams, utilized AI to create a synthetic version of Swift's voice, falsely claiming she was giving away free cookware sets to anyone who entered their personal information and shipping details. Le Creuset had nothing to do with the ads.

The incident is another example of the ease with which digital replicas of people are produced for deceptive purposes. In fact, as the South China Morning Post reported, just yesterday, a scammer created highly convincing videos of the CFO to scam an employee instructing him to wire $25 million out of the company’s bank accounts. This happened on a video call where the fraudster had created impersonations of not just the CFO but several other employees too to make the instructions more believable.

So what exactly are Deepfakes?

Deepfakes are synthetic media in which a person's likeness or voice is replaced with someone else's or artificially produced, making it appear as though they said or did things they never actually did. The technology behind deepfakes has advanced rapidly, thanks to machine learning and generative AI, making it increasingly difficult to distinguish between real and fake content.

The potential for harm is vast:

1. Individuals and Families: Deepfakes can be used to create non-consensual pornography, impersonate individuals for fraudulent purposes, or manipulate personal relationships, leading to psychological harm and social disruption.
2. Businesses and Brands: For businesses, deepfakes pose a risk to brand reputation, can lead to financial losses through stock manipulation or fraudulent activities, and undermine trust in the brand’s digital communications.
3. Society at Large: On a societal level, deepfakes can be used to spread misinformation, interfere with elections, and undermine public trust in media and institutions. President Biden was deepfaked just two weeks ago at the time of the New Hampshire primary. Former President Trump was deepfaked last summer.

The urgency to address the threat of deepfakes cannot be overstated. As we confront this challenge, a multifaceted approach will be required:

• Legislation and Regulation: Governments must move quickly (they haven’t so far) to enact stricter laws that address the creation and distribution of deepfakes, with clear penalties for those who misuse generative AI technologies. This also includes laws that protect victims and provide them with legal recourse. There seemed to be bipartisan support for such laws two weeks ago, but it now appears to have fallen off Congress’s radar.
• Technology Solutions: Tech companies need to invest in developing more sophisticated detection tools that can identify and flag deepfake content before it spreads. This includes collaboration between platforms to share information and strategies. Fake Catcher, DeepIdentify.ai, Sensity.ai, Optic, Sentinel.ai can all help.
• Public Awareness and Education: Raising awareness about the existence and dangers of deepfakes is crucial. People need to be educated on how to critically evaluate digital content and recognize potential deepfakes. It is always important to verify the provenance of the content. Meta’s announcement yesterday calling for an industry effort to label AI generated content is encouraging.
• Corporate Responsibility: Companies that develop AI and machine learning technologies must prioritize ethical considerations in their work, ensuring that safeguards are in place to prevent misuse including the creation of deepfakes. Microsoft CEO, Satya Nadella, upon hearing that the Taylor Swift deepfakes may have been created with Microsoft Designer said, “ We must act.” I obviously agree, they need toThe deepfake video from 2023 shows how convincing they can get

What can marketers do to fight deepfakes?

1. Implement Advanced Detection Technologies: Invest in cutting-edge technology that can detect deepfakes. This involves using AI and machine learning tools designed to identify inconsistencies or anomalies in videos and images that may not be perceptible to the human eye. Regularly monitoring content associated with the brand or endorsed celebrities can help in early detection of fraudulent materials. Check whether your agencies or internal technology and legal teams may have detection technologies in place.
2. Strengthen Legal Frameworks and Copyright Protection: Work with legal teams to ensure that copyright and intellectual property laws for your brand and intellectual property are enforced vigorously. This includes drafting clear contracts that address the unauthorized use of digital likeness and seeking legal remedies against perpetrators of deepfakes. Additionally, understand how the actors and the celebrities you work with are protected and whether they are susceptible to deepfakes or have been in the past.
3. Educate and Engage Your Employees and Customers: Create awareness among your internal teams and customers about the dangers of deepfakes. By informing your teams and customers how to spot fake content (services like AIorNot help) and encouraging them to report suspicious activities, brands can foster a community of vigilant and informed stakeholders. This also involves clear communication channels for reporting deepfakes and a responsive action plan to address concerns. The Swifties helped Taylor Swift on X, you need your own Swifties to aid your brand. And part of that includes having a crisis communication plan in place.
4. Leverage Digital Watermarking and Content Authentication: Implement digital watermarking and other content verification technologies to authenticate genuine brand content. By embedding invisible markers or using blockchain technology for digital certificates, brands can help audiences and platforms identify and verify the authenticity of the content they consume. As mentioned above, recent announcements by Meta and others are encouraging. Push your agency, publisher and platform partners to do more themselves.
5. Collaborate with Social Media Platforms and Tech Companies: Partner with social media platforms and tech companies to address deepfake content. This includes sharing best practices, developing standards for detecting and removing deepfakes, and advocating for platform policies that prevent the spread of AI-generated counterfeit content. Collaboration can also extend to sharing intelligence about new deepfake techniques and coordinating responses to emerging threats such as whether your advertising is appearing next to deepfakes. As I’ve mentioned in the past, as a marketer, you have greater influence on the technology ecosystem than you may realize - use that influence for the greater good.

The challenge posed by deepfakes is emblematic of the broader ethical and social dilemmas brought about by rapid deployment of generative AI. I would argue that we’re not giving the issue enough attention even after Taylor Swift was deepfaked twice in one month. I’m excited about the Super Bowl, the 49ers, the ads launching and how Taylor Swift is adding a whole new dimension to the day, however, we mustn’t forget the battle she and millions of others are fighting. She was horribly deepfaked and there’s nothing to prevent her or anyone else for that matter (including your brand) from being deepfaked unless we do more to fight the issue.

By confronting this issue head-on, we can mitigate its impact and protect the integrity of our brands and our own lives. Addressing the threat of deepfakes is not just about preventing harm; it's about preserving brand trust, authenticity, and the very fabric of our shared reality. After all, if you do not focus on fighting deepfakes for your brand, it may end up causing damage far greater than the cost of a Super Bowl ad.

Recent Savvy AI Articles

• People First in the AI Era (January 2024)
• Brace for Impact: GenAI's Role in 2024 (December 2023)
• Elon Musk Criticizes Marketers (December 2023)
• Research in the AI Era. Evolution or Extinction (November 2023)
• Thriving Professionally in the AI Era (November 2023)

What I’m reading

• India's ancient carpet weaving industry meets AI (BBC)
• 5 Best Books About Artificial Intelligence (The New York Times)
• Starbucks Uses AI-Powered Rewards to Boost Frequency (PYMNTS)
• Prompting Diverse Ideas: Increasing AI Idea Variance (UPENN)
• Why social media turned toxic and how we can fix it (FT.com)

Where I’m going and where I’ve been

It’s been an extremely busy January with trips to Indianapolis, Las Vegas and most recently Guadalajara, Mexico for corporate speaking engagements and board meetings. I am now planning to stay local for most of February as I focus on researching, interviewing and writing. And while, I’ve been local, I was invited to join Donnovan Andrews on his Doughnut Jar podcast. Enjoy the listen!

I would welcome the opportunity to speak at your event or educate your business teams on succeeding in the AI Era. Email me to discuss further.

What I’m writing about this week

I'm in the process of writing my third book, centered on artificial intelligence in the realms of business and marketing. This week, I'm editing a chapter on using generative AI to develop groundbreaking creative. Stay tuned for further updates and insights from the forthcoming book.