Teaching Students to Question the Machine

• Teaching students about whistleblowing—the willingness to act when something seems wrong—helps them develop the judgment, critical thinking, and ethical reasoning they will need to use AI responsibly.
• In a course at the Leeds School of Business at the University of Colorado Boulder, students discuss real-world corporate scandals before designing “speak-up” systems that provide employees with protected channels to escalate concerns.
• The exercise is meant to move students beyond viewing ethics as a set of rules to understanding that true accountability depends on company culture, employee independence, and organizational follow-through.

 
Artificial intelligence produces fast, confident, and plausible outputs. Unfortunately, organizations can put pressure on employees to accept these outputs without question—even when they are wrong.

Notably, this is not a problem specific to artificial intelligence. It is a problem of ethics.

We tend to frame AI ethics as a design challenge: better data, fairer models, clearer disclosures. Those things matter, but most of the AI failures students will encounter in their careers will not arrive as engineering problems. They will arrive as a colleague’s overconfidence, a manager’s shortcut, or an analyst’s number that no one wants to question.

That’s why I incorporate lessons from my research on whistleblowing into my leadership, business foundations, and business ethics courses at the Leeds School of Business at the University of Colorado Boulder. In fact, I would argue that one of the best ways to prepare students to use AI responsibly is to teach them to speak up when they see problems within their organizations.

In one assignment, I ask students to discuss a case in which an employee discovers that AI-generated data are fabricated; then, I have them design a system that would raise that concern to the company’s board of directors. The assignment helps them develop critical thinking, ethical reasoning, and a willingness to act when something seems wrong. Although AI is not the focus of the lesson, this exercise teaches them the skills they will need to use AI responsibly.

At Leeds, this approach reflects a broader philosophy: Rather than treating AI ethics as a standalone topic, we embed AI accountability into courses on leadership, governance, ethics, and social responsibility. By teaching students how to question information, challenge assumptions, and navigate organizational pressures, we prepare them to use emerging technologies with integrity long after today’s tools have evolved.

Bringing AI Accountability Into the Classroom

Across courses and learning activities, I follow a consistent approach: I present students with a real-world problem, ask them to work through it using established frameworks, and then assess their ability to connect individual decisions to broader organizational and policy structures.

Here’s the specific case I put in front of students: A junior analyst examines a financial analysis that her manager has produced with the company’s preferred generative AI tool. The analysis is already circulating as fact. But when the junior analyst checks the underlying figures, she realizes the model fabricated them.

By teaching students how to question information, challenge assumptions, and navigate organizational pressures, we prepare them to use emerging technologies with integrity long after today’s tools have evolved.

She raises her concerns quietly with her manager, and his reaction sets the stakes: If this surfaces, he loses his job. What should she do next? And what should her organization have done, long before this moment, to ensure she is not left facing a choice between protecting her career and doing the right thing?

To respond effectively in those moments, students will need to do more than memorize a policy. They will need to apply durable skills such as critical thinking, ethical reasoning, the ability to recognize when something is wrong, and the knowledge of how to respond.

Designing Systems for Speaking Up

Students will understand their ethical responsibilities more fully when they examine how recent corporate scandals came to light. For example, this semester, I am giving a guest lecture on whistleblowing and governance in the course Corporate Boards in Action. In it, I begin by defining whistleblowing using Janet Near and Marcia Miceli’s classic 1985 definition: the disclosure by current or former members of an organization of practices they believe to be illegal, unethical, or illegitimate to parties who can take action.

We then examine Theranos, where board members, auditors, and investors all failed to detect the deep fraud the company’s leaders had perpetrated. For a time, employees who raised concerns internally served as the company’s only effective accountability mechanism. In many ways, whistleblowing stepped in where governance had failed.

Next, I present students with the case described above involving the junior analyst and ask them to complete two tasks. First, they must determine how the analyst should escalate her concern—through which channels, to whom, and with what protections—so that it reaches the board rather than dying on her manager’s desk. Second, they work in small groups to design a speak-up system that would make that escalation possible in a real organization and explain what structures, processes, and cultural supports must exist, beyond a written policy, for the system to work.

Finally, each group presents its system design to the class, and a structured debrief draws out the reasoning behind each choice. Students are evaluated not only on the quality of their proposed governance structures, but also on their ability to justify decisions, identify tradeoffs, evaluate which solutions are most likely to succeed, and connect their recommendations to broader questions of organizational accountability.

What Kinds of Systems Work?

It is during the task of designing the speak-up system that much of the learning happens. Students quickly discover that a reporting channel on paper is not the same as one people will actually use. Employees are unlikely to speak up if they fear retaliation, lack confidence in the process, or believe their concerns will be ignored.

To establish systems that are both credible and effective, students draw on established frameworks, including the EU Whistleblowing Directive (2019/1937); the ISO 37002:2021 guidance on whistleblowing management systems; and practitioner recommendations from The Whistleblowing Guide by Kate Kenny, Wim Vandekerckhove, and Marianna Fotaki.

Students must consider issues such as how and whether to ensure anonymity, who should investigate when the implicated individual is a senior leader, how to protect employees from retaliation, and why a culture that punishes the messenger can render even the strongest policy ineffective.

In designing their speak-up systems, students are asked to consider three core elements: trust, impartiality, and protection.

So, what does a system people actually use look like? In designing their systems, students are asked to consider three core elements identified in research and practice: trust, impartiality, and protection. Each one translates into a concrete design choice.

Trust requires more than a reporting channel on paper. Employees need multiple ways to raise concerns, including a path that does not run through the very manager they are worried about. They also need confidence that reports will be acknowledged and taken seriously.

Impartiality means having a system in place where concerns—especially those involving senior leaders—are reviewed by someone independent of the situation, while also ensuring fair treatment of the person named in a report.

Protection requires shielding the reporter from retaliation in all its forms—not only dismissal, but also quieter detriments such as exclusion, a stalled promotion, or the cold shoulder. Such protection is difficult to implement, but it is central to effective speak-up systems. That protection also should be visible to others, because the first unpunished reprisal teaches everyone else to stay quiet.

As they work, students must confront real trade-offs. Anonymity may protect employees but complicates investigation and follow-up. A hotline is inexpensive but useless if no one trusts who sits on the other end. A policy can promise nonretaliation, but only leadership behavior makes that promise believable.

The goal of the exercise is to move students beyond viewing ethics as a set of rules and toward understanding that accountability depends on company culture, employee independence, and organizational follow-through. That is also the most portable part of the exercise for other instructors: Hand students a case of plausible wrongdoing, give them a real framework to adapt, and let the trade-offs teach the lesson.

The Three Skills That Students Develop

As they work through the exercise, students develop three capabilities that are essential for using AI responsibly:

Independent judgment. Generative AI models produce fluent, confident, and plausible outputs, while organizational hierarchies often create pressure to defer to authority. Yet both models can be wrong—and sometimes wrong in the same direction. In the case study, for instance, the analyst must trust her own conscience and reasoning even when her conclusions conflict with both the machine and her manager. Employees need this skill when an algorithm’s recommendation appears authoritative but is not sound.

A deeper understanding of AI’s ethical stakes. Abstract harms—such as biased hiring tools, opaque credit decisions, and fabricated outputs passed off as analysis—become more concrete when students trace how these harms would surface inside a real company. Discussions of real-world scenarios show students that many of the most important AI-accountability moments have come from insiders, not auditors. Students leave with a clearer understanding that transparency and fairness often depend on someone inside an organization noticing a problem and being willing to raise it.

Governance literacy. Students learn that in their future careers they will need fulfill two roles. In the first, they will be employees who raise concerns. In the second, they will be leaders who build cultures where clear internal reporting routes exist, ethics oversight occurs at the board level, and strong protections from retaliation are in place. This reframes whistleblowing from an act of disloyalty into an early-warning system that well-run organizations want to have.

Reinforcing the Public-Policy Dimension

I integrate such ethics-based themes throughout my core course as well. In Business Ethics and Social Responsibility, we examine major corporate scandals at companies such as Enron, WorldCom, and Wells Fargo. For the latter, some students write ethical analyses on the Wells Fargo cross-selling example; others debate the ethicality of the actions proposed to address the scandal. We do not treat these as ancient history, but as examples of how routinely people in business breach ethical standards. Nearly every one of these collapses came to light only after a whistleblower from inside the organization brought attention to wrongdoing.

We return to the topic of how companies and governments respond to scandal when we study collective action and public policy; we look especially at how individuals and groups hold powerful organizations to account and how that pressure can become codified into law. For the most part, the EU Whistleblowing Directive, the Sarbanes-Oxley Act of 2002, and the Dodd-Frank Act of 2010 exist because people spoke up and lawmakers responded.

Students are assessed on this connection through a quiz on whistleblowing as a mechanism of collective action and policy change. This assessment keeps the civic dimension from getting lost behind the individual drama of any single case. Taken together, the written analyses, debates, and quiz create a layered picture: Students practice individual ethical reasoning and then situate it within the organizational and legal structures that make speaking up possible or impossible.

I cannot teach students every rule for every tool they will eventually use. What I can do is cultivate their courage to act when something is wrong.

This work extends naturally to environmental accountability, where so much depends on insiders willing to challenge claims. For instance, in 2021, DWS, the asset-management arm of Deutsche Bank, was accused of overstating its environmental and social investing. That case began when DWS’s former head of sustainability, Desiree Fixler, went public with the allegation that the firm had painted a rosier picture than reality supported. The fallout was not minor: U.S. regulators imposed a 19 million USD fine in 2023, and Frankfurt prosecutors added a 25 million EUR penalty in 2025.

Greenwashing, like fabricated financial statements, often comes to light because someone inside an organization decides the gap between the claim and the reality is too large to ignore. That is exactly the judgment I want students to practice.

Why It Matters Now

AI might be a new technology, but it raises the stakes on two old skills: recognizing when something is wrong and designing the structures that allow people to speak up.

I cannot teach students every rule for every tool they will eventually use; the tools will change faster than any syllabus. What I can do is teach them to use AI responsibly by cultivating their judgment, critical thinking, ethical foundations, and courage to act when something is wrong.

Business schools can ensure that graduates know not simply how to use powerful tools, but how to question them. Teaching whistleblowing is one of the most effective ways to build those capabilities. This is true even though, and partly because, AI itself is never the lesson.