by On May 18, Starbucks Corporation woke up to a reputational crisis its own teams had no hand in creating. Thousands of miles away, its South Korean licensee, Shinsegae Group, had launched a tumbler promotion called “Tank Day,” on the 46th anniversary of a military massacre, using a slogan that echoed one of the country’s most infamous political cover-ups. The campaign was canceled within hours. The CEO of Starbucks Korea was fired. Police opened a criminal investigation. South Korea’s president weighed in publicly. By the time Shinsegae Group Chairman Chung Yong-jin delivered a nationally televised apology, the Starbucks name had spent days at the center of a national controversy that Starbucks Corporation had no operational role in producing.
The case raises a question not only for global brands and franchisors. The question is also for private equity firms, institutional investors, etc., to answer before the crisis arrives: who owns the reputational risk when the legal structure separates ownership from operations?
What Happened in South Korea
Starbucks Korea’s marketing team used an AI tool to generate campaign concepts for a new stainless-steel tumbler it called the “SS Tank.” The tool produced ideas; the team selected one. They chose May 18 as “Tank Day” — a date that every Korean adult recognizes as the anniversary of the 1980 Gwangju Democratization Movement, in which the military government deployed tanks and paratroopers to suppress a pro-democracy uprising, killing hundreds. The promotional slogan, “Thwack it on the table,” compounded the damage: it directly echoed the infamous 1987 police statement used to cover up the death by torture while in custody of student activist Park Jong-chol.
According to a statement by Shinsegae Group, the campaign passed through four to five layers of internal approval. Some of the managers who signed off never opened the email attachments showing the actual marketing material.
The speed of the crisis that followed reflects the depth of the failure that preceded it. Shinsegae canceled the promotion within hours, terminated the Starbucks Korea CEO, cooperated with a police investigation, closed every Starbucks store nationwide for mandatory history and sensitivity training, and had its group chairman bow before national television cameras in a formal apology. The response, once it arrived, was largely competent. The problem is that it should never have been necessary. Competent crisis response is not a substitute for crisis prevention, and the organization that executed a credible response within days clearly had capable people who were simply absent from the process that mattered most.
The Structure Behind the Story
Starbucks Corporation transferred direct ownership of its Korean operation to Shinsegae’s E-Mart subsidiary in 2021. It licenses the brand; it does not hold equity. That legal distinction carries real consequences in a reputational crisis, because public perception does not follow the ownership org chart. Consumers, media, and regulators see the Starbucks name. The reputational damage travels upstream to the highest level of the brand regardless of who made the decision.
This structural vulnerability is not unique to brand licensing. It runs through several of the most common corporate arrangements in global business. A private equity firm that holds a controlling stake in a portfolio company does not run its day-to-day operations, but its name surfaces in the headlines when a portfolio company crisis becomes public. A parent company with a partially integrated subsidiary after an acquisition inherits the subsidiary’s cultural blind spots along with its revenue. Joint venture partners bring their brand equity to a shared enterprise without retaining full control over each other’s decisions. In each case, the architecture of the relationship creates distance between named accountability and operational reality.
The instinct to preserve that distance is often sound. PE firms that impose heavy operational control on portfolio companies risk damaging the culture and values that made the investment thesis work. Parent companies that absorb subsidiaries too aggressively create the very disruption they sought to avoid. The reverse is equally true: operational decisions pushed down from a PE firm or parent company without adequate sensitivity to the specific market or stakeholder environment of the operating entity carry their own reputational risks. The governance question is not about control. It is about where the communications function sits in the relationship, and whether it has visibility in both directions.
The Accountability Gap AI Is Making Wider
The Starbucks Korea case is a governance failure with two distinct layers, and the second is newer and less well understood than the first.
The first layer is familiar: an approval chain that was nominal rather than functional. Managers who sign off without reviewing the material are not providing oversight: they are providing paperwork. Four to five approval layers that catch nothing are, in practice, no approval layer at all. This kind of insular process, internally coherent and externally blind, is a recurring feature of corporate crises that should have been preventable.
The second layer is more recent. The campaign originated from AI-generated ideation. The tool produced a concept that was linguistically plausible and creatively coherent. But, short of posing a prompt into the AI system that sought to find this underlying faultline, had no way of knowing what May 18 means to a Korean adult. That is not a failure of the technology: it is a failure of how the technology was deployed. AI used to generate decisions, rather than to challenge and stress-test them, removes the critical friction that catches culturally or historically illiterate concepts before they reach production. A properly designed human-AI workflow runs in both directions: AI surfaces ideas, and humans (or AI prompted differently) interrogate those ideas for risk. The question “what could go wrong with this in this market?” should be a standard step in any AI-assisted creative process. At Starbucks Korea, nobody appears to have asked it. The benefits of using AI are largely in amplifying human critical reasoning, not in failing to use it.
For franchisors, investors, and parent companies, this creates a new category of reputational exposure. If your operating entity is using AI in its marketing and decision-making processes and you have no visibility into how, you are carrying a risk that almost certainly does not appear anywhere in your licensing or investment agreement.
What the Legal Agreement Doesn’t Cover
Most licensing, franchise, and investment agreements are thorough on the dimensions their drafters anticipated: product standards, financial covenants, audit rights, exit mechanisms. Very few address the communications governance questions that a crisis makes urgent.
Crisis escalation rights (who has the standing to intervene in the other party’s communications at what threshold, and on what timeline) are almost never defined. When a crisis breaks, both parties improvise the governance in real time while simultaneously managing the crisis itself. Cultural and market risk protocols are rarer still; the standard assumption is that local management knows the local market. The Starbucks Korea case demonstrates precisely how that assumption fails. And AI governance is genuinely new territory: most agreements predate the widespread use of AI in marketing and operations, and no standard clause yet addresses how AI-assisted decision-making should be reviewed, by whom, or with what escalation rights for the senior partner in the relationship.
The underlying principle is straightforward: communications governance is a structural question, not an operational one. It belongs in the agreement, not in the aftermath of the crisis the agreement failed to anticipate.
Building the Architecture Before It’s Needed
The organizations that manage cross-structural reputational risk well share a common characteristic: they treat it as a design problem rather than a response problem. That requires two things working together.
The first is systematic pre-event analysis. At Montieth & Company, our proprietary predictive intelligence tools enable organizations to stress-test decisions (campaigns, market entries, executive appointments, structural changes, transactions, legal proceedings) against reputational, cultural, and stakeholder risk before those decisions become commitments. Applied at the licensing or investment stage, that kind of analysis surfaces the reputational and perception gaps that standard due diligence misses. Applied operationally, it provides the critical friction that the Starbucks Korea approval chain entirely lacked.
The second is in-market human judgment. Intelligence tools identify patterns and surface risk scenarios. They do not know what May 18 means in Gwangju. Advisors with genuine local knowledge and established stakeholder relationships do — and they are most valuable precisely at the moment before a decision is made, not after a headline has been written. For global brands and investors operating across multiple markets, geographic depth is not a supplementary resource. It is a core part of the reputational governance infrastructure.
The Starbucks Korea case will become a reference point in how global brands think about licensing, investment structures, and reputational accountability. The lesson it offers is not primarily about crisis response. Shinsegae’s response, once mobilized, demonstrated real capability. The lesson is about the governance architecture that should have made the response unnecessary. The organizations that take that lesson seriously will be the ones that have already answered the question before the next crisis asks them.
Learn more about how Montieth & Company approaches crisis and issues management and corporate and financial communications.