Can you trust an unethical person? Ask the question and you’ve already answered it. Trust evaporates in the absence of integrity. The most memorable lesson my father taught me in my short fourteen years with him came during an after-dinner walk around the university during his last month.
At one point, he stopped, put his hand on my shoulder and fixed his eyes on mine. “Living a life without ethical principles is like a vessel with a rounded bottom,” he said. “It wobbles aimlessly and never stands firm, lacking conviction and character, and eventually tips over.” It took me years to fully grasp the depth of his wisdom. Yet his parting insight has spared me lawsuits, betrayal and heartache more times than I can count.
Go into business with someone willing to cut ethical corners, and you may find yourself on the wrong end of the law. Hire the win-at-all-costs hotshot, and you’ll spend your days pulling knives out of the backs of team members. Tie yourself to a deceitful ally, and feel the rope burn your hands—and reputation. The same lesson of character my father shared now applies to our algorithm-driven business world.
Layer in AI, and those ethical wobbles become earthquakes. Deploy facial recognition for “security”, and civil liberties hang in the balance. Optimize solely for engagement, and you’re implicitly choosing clicks over well-being. Just ask Amazon why it scrapped its experimental AI hiring tool. Or Meta why its engagement algorithm drew the US Congress’s glaring gaze. These weren’t failures of technology. They were failures of judgement. When data and code drive decisions without human oversight, bias doesn’t disappear; it gets automated.
Leaders love to believe that AI is neutral and objective. It isn’t. It’s a reflection of an organization’s ethics. These aren’t technical decisions. They’re moral ones. If you want your team and company to stand strong, you can’t outsource ethics to AI. What my father taught me about character applies to code, too. Why? Because AI evolves faster than Moore’s Law. Every month brings new capabilities, new risks, new temptations. Either you proactively shape your organization’s code of ethics—or you’ll inherit one written by accident, crisis or silence. An AI code of ethics isn’t a third-tier priority. It’s urgent. And with every model your company adopts, the threats to customer and employee trust grow.
In a room of senior executives, I once asked, “Who owns AI ethics in your company?” Someone pointed to legal. Another to compliance. One said, “We have a GenAI task force.” I replied, “Wrong. You own AI ethics in your company. And if you don’t steer your AI ethics, the machine will.” AI ethics is not a department. It’s a leadership decision. Do your people know exactly what data sets and documents they may—and may never—feed into an LLM? Have you built incident-response and AI audit trails as robust as your data breach protocols? Do your team members know which AI-generated concepts are legal—or are you allowing them to wander in the dark into a copyright landmine? If you don’t have the answers, start looking for them now. Regulators are catching up. Users are catching on. Talent is choosing companies that lead with values.
If your AI causes harm, apologies won’t suffice. If your AI discriminates, audits won’t save you. And if your people can’t trust how AI is being used, they will leave. You don’t get credit for post-disaster clean-up. Leadership is forethought. It’s not only about preventing your company from imploding.
An AI code of ethics is also about enshrining your organization’s values into the way your people build and deploy AI. At Sampark, we drew a bright line early. We had opportunities to vacuum up more data. To track learning behaviours with specificity. To correlate content usage with household demographics. We even developed a tool to detect children’s learning difficulties by tracking their facial expressions through smartphone cameras. Granular. Data-rich. Enticing. But it felt wrong. We said no. Why? Because our teachers and children trusted us. And that trust was more valuable than any insight we might gain from surveillance. We didn’t wait for an ethics scandal. We pre-empted it by making sure our tools reflected our values. That’s what every organization must do now—before scale turns an ethical wobble into a whirlwind.
The question, of course, is how can you spark an AI ethics code that prevents harm and propels your values? Start right now by taking three actions.
First, embed five layers into your ethical stack. Begin constructing your ethical stack by focusing on the 5Ds. These are i) Data: Where did it come from? Whose voice is missing? ii) Design: What’s the system optimized to do—and not do? iii) Deployment: Who is impacted? What’s the feedback loop? iv) Decision-Making: Who can override it? How are trade-offs logged? v) Disclosure: Will the user know it is AI? Can they opt out? Honing the 5Ds will place your company’s character on firm ethical footing.
Second, teach ethics from day one—and keep teaching. From day one, new hires must be taught your company’s AI ethics code during onboarding so that guard rails are in place before customer or corporate data ever touch their hands. Then all team members should see your AI ethics code reflected in daily workflows. When your team learns early and sees ethics modelled daily at every point on the decision-chain, trust becomes the default setting—not PR clean-up after an ethics lapse turns into a calamity.
Third, hold quarterly AI safety drills. Every three months, gather legal, HR, engineers and trusted users to stress-test the system before it violates trust and your code of ethics. Hunt for privacy leaks, lurking bias or ways your platforms could be contorted to create harm. Putting people first, machines second means testing hard now so no one gets hurt later.
My father’s words still echo: without ethics, we wobble. That’s doubly true in the age of AI. It will scale your company’s soul—or its shadows. I’ve seen organizations wobble like that rounded-bottom vessel my father described, until one ethical misstep tipped them over.
Every model you build writes a bit of your legacy. Every shortcut you take today will be the scandal someone else investigates tomorrow. Don’t ask: Do we need an ethics policy? Ask: Have we earned the trust to deploy this technology at all? If the answer isn’t yes, then pause. Revisit the code—not just the algorithm, but the human one. Refine it now. Before the machine decides what you really believe.
Edited excerpt from Humans First, Machines Second, by Vineet Nayar, published with permission from Penguin Business.
Write to us at feedback@livemint.com