Why Text-to-Image AI Requires a New Branding Mindset

In the world of marketing, brand identity is the castle that guards a company’s reputation, with a moat that keeps the competition at bay. One doesn’t need to look any further than T-Mobile’s signature magenta hue (226 red, 0 green, 116 blue) to realize how fiercely this castle is protected. Use a similar shade, and you might find a cease-and-desist letter in your mailbox.1

But what if the very ramparts that have protected brands could hinder their future in the age of generative AI?

In one notable advertising campaign, Heinz turned to OpenAI’s DALL-E generator to create ketchup-inspired images. The images weren’t perfect in terms of the dimensions or colors, but they were undeniably Heinz.2 This shows that even AI models recognize that the Heinz brand has top-of-mind awareness. An old-school brand purist might have raised an eyebrow at the way the brand was represented, but the campaign created undeniable buzz, with 850 million earned impressions around the world.

We’ve experienced the transformative potential of generative AI on branding in our own classrooms. Generative AI breathed life into King’s College London’s mascot, a fiery-red lion named Reggie, as part of the school’s centennial celebration. Thanks to AI tools, the mascot became more than just a static brand symbol: In classroom sessions, Reggie became animated and “alive,” leading some discussions and even sorting through clothes and adjusting prices as we discussed merchandise management. The result: Bland stock images gave way to a beloved mascot that resonated with our students and elicited overwhelmingly positive reactions. But this is just the beginning of these tools’ potential for brands.

Creativity Versus Control: Choose Wisely

Brands ready to shake off the old, protective mindset will find that generative AI opens new doors to cocreation. Text-to-image AI models, such as OpenAI’s DALL-E, eliminate the barriers for everyday consumers to generate visual designs, allowing unprecedented levels of personal connection and customization.

In Coca-Cola’s groundbreaking initiative “Create Real Magic,” consumers were invited to unleash their creativity by designing advertisements using ChatGPT and DALL-E.3 The winning entries won’t merely be lauded; they will soon light up the billboards of some of the world’s most iconic advertising stages, including New York’s Times Square and London’s Piccadilly Circus.

Our research, as well as that of others, shows that these enriched connections don’t just affect the people directly involved; they ripple outward, strengthening relationships and enhancing purchase intentions across the broader market.4

Despite these exciting prospects, many brands are reluctant to leave the sidelines, likely held back by fears of losing control. The age-old approach of strictly safeguarding brand identity and symbols could make them lagging adopters of text-to-image models.

Today’s reality is that absolute control over AI technology remains a pipe dream. While generative AI will get better at rendering consistent styles, it is unable to adhere to the rigid guidelines demanded by many brands today. As seen in Heinz’s campaign, while the technology replicated the familiar bottle shape and color, it fell short of reproducing the exact logo or brand name. Brands banking on a “ready” version of this technology might be in for a long wait. Perhaps more importantly, a myopic focus on control could prevent brands from fully unlocking the doors to consumer engagement that generative AI offers.

Brand Managers Need to Be Enablers

The path forward requires a mindset shift: Brand managers must evolve from enforcers and guardians to enablers. It is about focusing on broader innovation and connection opportunities instead of obsessing over minutiae like exact color values.

GoFundMe’s “Help Changes Everything” campaign made new consumer connections by using text-to-image models to engage with donors in the most personalized way possible. Rather than producing typical year-end content, GoFundMe used tools like DALL-E and Stable Diffusion to produce a series of AI-generated street mural-style images showcasing all of the fundraising campaigns and donors of the past year. These individual pieces were then compiled into a captivating video that resonated strongly with audiences and received widespread acclaim.5

Brand managers who are open to such opportunities can leverage text-to-image models to transcend traditional communication boundaries and achieve previously unattainable levels of personalization and authenticity.

Of course, risk appetites differ. What works for a daring brand like Monster Energy might not for a conservative institution. But even in highly regulated and conservative settings, it is possible to experiment in a way that constrains risks and costs. The risks and potential negative consequences of one-off campaigns or small-scale projects are limited and thus are ideal opportunities for brand exploration. Our endeavor to animate our university’s mascot serves as a case in point.

Experiment Safely: Lessons From Coca-Cola

This isn’t a call for anarchy. Brands can explore the possibilities of generative AI while keeping within their comfort zones and established boundaries. Coca-Cola’s protective measures include using its own platform, maintaining oversight over content displayed on its site, and providing participants with Coca-Cola-themed templates, all to retain a degree of control over brand portrayal.

Coca-Cola’s initiative was done in partnership with a text-to-image service alliance of Bain and OpenAI, which helped it switch from a reactive stance to a proactive one.6 Providers that offer brands such as Coca-Cola security and safeguards will likely emerge as preferred partners.

Of course, brand managers must guard against potential misuses and walk a tightrope between creative freedom and control. But let’s not forget: Brand perceptions have always been coauthored with consumers and have depended on an intricate two-way relationship between brands’ actions and consumers’ reactions. Indeed, consumers have long been instrumental in molding brand perceptions, whether it’s through word-of-mouth endorsements or scathing reviews. Generative AI only intensifies this dynamic, empowering the consumer to wear various hats: They can play the role of designer, cinematographer, or even software developer. In the age of generative AI, active listening, consistent monitoring of consumer sentiments, and an agile approach to recalibrate strategies based on real-time feedback and emerging trends become even more important.

While brands cannot control every narrative twist, they can guide the dialogue more proactively. One effective approach is to cultivate environments where consumers can connect with one another and the brand and contribute to the brand story within a framework of transparent rules, incentives, and recognition. Coca-Cola’s generative AI campaign, for example, clearly communicates the value it places on consumer input — by rewarding top contributions with highly visible advertising placements, as well as a chance to be selected for a trip to the Coca-Cola headquarters to participate in a content-creation workshop. The brand has invested in a specialized digital platform tailored to promote innovation and deepen consumer relationships while at the same time mitigating risks and maintaining alignment with its core values and identity.

Addressing Ethical and Legal Concerns

Brand managers also need to be vigilant about ethical and legal issues surrounding text-to-image models. As lawsuits entangle key AI tool companies, staying abreast of these developments around ownership and intellectual property (IP) rights becomes essential.7 Brands are legitimately concerned about legal implications, but where challenges arise, innovation follows.

Adobe offers one innovative example of how to ease customer concerns regarding IP issues. For its Firefly generative AI tool, which lets people create or enhance images using simple prompts, Adobe trained the AI models on stock images, openly licensed content, and public-domain materials free from copyright constraints. That’s a forward-thinking approach aimed at addressing customers’ concerns.

Another promising approach is collaborative IP creation agreements that focus on mutually beneficial relationships between AI developers, creators, and brands. A case in point is Shutterstock’s Contributor Fund, which compensates creators when their images are employed in training generative AI models.8

Brand managers should opt for partnerships with platforms that have effectively mitigated IP risks. Brand managers could also consider adding specific protections in their contracts, where the platforms confirm that they have properly licensed training data and provide guarantees for any potential IP infringement cases against the brand. For generative AI developers, this serves as a compelling incentive to develop models that respect the copyrights of creators and positions them as desirable collaborators for brands.

All in all, text-to-image generative AI models are tearing down the walls of the traditional branding castle, creating a new reality ripe with opportunities, risks, and challenges. As we navigate this transformation, every brand manager must confront a vital question: Will your brand be a builder of bridges or barricades in this new era?