Foundational Ethical Principles in Technology

Imagine a health app that promises to keep your medical records perfectly safe—so safe that only a tech expert can log in. Your grandmother, who needs it most, can’t use it. That conflict between protection and real-world fairness sits at the centre of every technology we build. This chapter explains the main ethical ideas that help us handle such tough choices, so we can design tools that truly serve everyone.

The Big Picture#

Every technology, from a simple smartphone app to a nationwide surveillance system, makes choices about who benefits, who gets harmed, and who is left out. These choices aren’t just technical—they’re deeply ethical. In this chapter we introduce four key principles—beneficence, nonmaleficence, justice, and efficiency—that act as a compass for responsible design. We then explore what happens when these principles conflict, especially the classic struggle between very tight security and real accessibility. By the end, you’ll see there is no perfect answer; the goal is to weigh trade-offs carefully and keep human dignity at the centre.

The Four Pillars: Beneficence, Nonmaleficence, Justice, and Efficiency#

When we set out to create a new technology, we almost always have a positive aim in mind: to help people communicate faster, to improve health, to make daily tasks easier. But good intentions are not enough. Ethics gives us a shared language to examine our designs from multiple angles.

Beneficence: The duty to actively do good—to design technology that promotes well-being and adds value to people’s lives.

Think of a fitness tracker that nudges you to move more. Its whole purpose is beneficence: it aims to improve your health. But we have to ask: whose well-being? Does the tracker accidentally shame users who can’t meet step goals because of a disability? Beneficence pushes us to look beyond the average user and ask who is actually being helped.

Nonmaleficence: The duty to avoid causing harm—often summed up as “first, do no harm.”

This is the principle that makes us pause before releasing a feature. A social media platform might introduce an autoplay video feature to increase engagement (a benefit), but if it leads to information overload, anxiety, or addiction, it violates nonmaleficence. In technology, harm can be physical (a self-driving car’s sensor failing), psychological (a recommendation algorithm amplifying content about eating disorders), or social (a facial recognition system that misidentifies people and leads to wrongful arrests). Nonmaleficence demands that we foresee and reduce these harms.

Justice: The duty to distribute benefits and burdens fairly across all groups, without discrimination or unfair exclusion.

Justice asks: who gets access to this technology, and who bears its costs? If a city installs air-quality sensors only in wealthy neighbourhoods, the data will guide policies that benefit those residents while ignoring pollution in poorer areas. Justice demands that we correct such imbalances. It includes fairness in how algorithms treat different groups, how privacy risks are distributed, and how economic opportunities from automation are shared.

Efficiency: The goal of achieving maximum benefit with minimum waste of resources, time, or effort.

Efficiency is often the easiest principle for engineers to love. We want code that runs fast, batteries that last long, and systems that scale cheaply. But efficiency alone can be a bully. A perfectly efficient facial recognition system that works only for light-skinned faces is efficient for a narrow slice of the population, but it goes against justice and nonmaleficence for everyone else. Efficiency should support the other principles, not push them aside.

These four principles are not a checklist to tick off. They are lenses that we switch between while making decisions. A design that maximises beneficence might accidentally create new harms (nonmaleficence), or it might be so expensive that only the rich can afford it (justice). The real art of ethical technology is navigating the tensions among them.

📝 Section Recap: Beneficence, nonmaleficence, justice, and efficiency are the four core ethical principles that guide technology design. They often pull in different directions, so responsible design requires constant balancing.

When Good Goals Collide: Security vs. Accessibility#

One of the most common and painful trade-offs pits strong security against accessibility for vulnerable groups. The dilemma is simple: the more barriers you put up to keep attackers out, the harder it gets for real users—especially those who aren’t tech savvy, have disabilities, or use older devices—to log in.

Imagine a banking app that requires two-factor authentication, a complex password changed every month, and a fingerprint scan. For a young, tech-savvy user with a modern phone, this is a minor annoyance. For an elderly person with arthritis who struggles to type, or for a visually impaired person using a screen reader that can’t handle the login sequence, those same security steps become a wall. The bank has achieved high nonmaleficence (protecting accounts from fraud) and efficiency (automated security), but at the cost of justice: it has locked out some customers from managing their own money.

This is not a rare, edge-case scenario. During the COVID-19 pandemic, many governments launched digital vaccination certificates. In some countries, these certificates could only be stored in mobile apps that needed the newest phone software and a reliable internet connection. People without smartphones—disproportionately the elderly, the homeless, and low-income individuals—were locked out of travel, dining, and even employment. The system was efficient and secure, but it made existing inequalities worse.

Accessibility: The practice of designing products so that people with a wide range of abilities, including those with disabilities, can use them easily and effectively.

Accessibility is not a niche add-on; it is a direct expression of justice. When we talk about security versus accessibility, we are really talking about a tension between nonmaleficence (preventing data breaches, identity theft) and justice (ensuring equal access). There is no one-size-fits-all rule to resolve this tension. Instead, we must ask questions that fit the situation: What’s the worst that could happen if security fails? Who gets hurt when a group is locked out? Can we give users different security levels so they can pick what works for them?

For example, a health portal might allow patients to set a simple four-digit PIN if they also use a fingerprint or face scan on their device, while offering more complex passwords for those who want them. The key is to involve the affected communities in the design process—a practice called co‑design (designing together with the people who will use the system)—so that their real needs shape the trade-off, not just engineers’ guesses.

📝 Section Recap: Strong security can accidentally lock out people with disabilities, older adults, and those with limited tech access. Balancing nonmaleficence and justice asks for flexible, user‑centred design rather than one-size-fits-all barriers.

Unintended Consequences: How Security Measures Can Backfire#

When security measures become too hard to follow, people don’t just give up—they find workarounds. And those workarounds often make the system less secure than if the security had been easier to use in the first place. This is what we call a perverse outcome: an action meant to increase safety that actually decreases it.

Perverse outcome: an action intended to make things safer that ends up making them less safe.

Consider the rule that bank cards must have a PIN. To protect against theft, the system requires that the PIN be memorised and never written down. In practice, many people—especially those juggling dozens of passwords—write the PIN on a slip of paper and keep it in their wallet, right next to the card. Others use the same obvious PIN for everything. The security policy, applied too strictly, asked more of people’s memory than they could manage, so they bypassed it in the riskiest way. The goal was to do no harm, but the result made them more vulnerable.

This pattern appears everywhere. A hospital’s digital health record system requires a 16-character password that changes every 30 days. Nurses, racing to save lives, start leaving their passwords on sticky notes attached to monitors. The security measure, designed to protect patient privacy, now exposes it to anyone walking by. The burden fell heavily on the nurses, who already face immense time pressure—a redistribution of risk from the hospital’s legal liability onto the frontline staff.

Redistribution of risk: when a design shifts risk from one group to another instead of eliminating it entirely.

Redistribution of risk is a crucial concept. When we design a system, we are not erasing risk; we are moving it around. A strict password policy may shift the risk of a data breach from the IT department to the individual user, who now must remember a complex string of characters. If that user has dyslexia or memory issues, the risk shifts again—onto their ability to access essential services. Income level plays a role too. A wealthy person can afford a password manager and a new phone with fingerprint or face recognition; a low-income person relying on a shared library computer cannot. The security design, however neutral it looks, redistributes the burden along lines of income and ability.

This does not mean we should abandon security. It means we must map out where the burden lands and ask whether that distribution is fair. A security measure that protects a corporation’s reputation while making life harder for its most vulnerable customers is unfair. Sometimes the most ethical choice is to accept a slightly higher chance of a breach in order to avoid real, everyday harm to disadvantaged groups.

📝 Section Recap: Overly strict security often triggers workarounds that weaken protection. Risk is redistributed, not eliminated, and the burden frequently falls on those least able to bear it—people with disabilities, the elderly, and the economically disadvantaged.

The Ethical Tightrope: Perfect Data Protection vs. Justice and Safety#

Some people argue that we should always aim for the strongest possible data protection, scrambling everything so that not even the service provider can read it. On the surface, this sounds like a triumph of nonmaleficence and beneficence: no one can misuse data they cannot access. But absolute protection can conflict with justice and even with overall safety.

Imagine a domestic violence shelter that runs a secure messaging platform for survivors. To protect them from abusers, the platform uses end-to-end encryption and automatically deletes messages after they are read. One survivor uses the platform to tell staff that her abuser has found her new address and is on his way. She then loses internet access. The shelter staff, unable to retrieve the message because of the encryption and auto-deletion policy, cannot act in time. The system perfectly protected her privacy, but it also prevented a life-saving intervention. The ethical trade-off is clear: perfect data protection in this case went against helping her and even against avoiding harm, because it blocked the very people who could protect her.

This dilemma also surfaces in public health. During a disease outbreak, contact-tracing apps can slow the spread. If the app is designed to be perfectly private—storing no location data, sharing no identifiers—it may be less effective at warning people who were exposed. A slightly less private design that shares anonymised records of who you’ve been near with health authorities could save lives. Here, the trade-off is between individual privacy (a piece of justice and nonmaleficence) and public safety (a form of beneficence). There is no one right answer; the ethical task is to make the trade-off open and to involve the affected communities in the decision.

The concept of proportionality helps. Proportionality means that any invasion of privacy or reduction of security should be no greater than what is necessary to achieve a real, important goal. If a less intrusive measure could achieve the same benefit, we should choose it. For example, a contact-tracing app might use Bluetooth signals that automatically expire after 14 days rather than GPS tracking that creates a permanent map of your movements. The ethical designer asks: What is the least harmful way to achieve the good we seek?

Trade-offs like these are not bugs in the system; they are features of a world with limited resources and competing values. The goal is not to eliminate tension but to navigate it with honesty, humility, and a commitment to those who are most affected.

📝 Section Recap: Perfect data protection can sometimes prevent life-saving actions or deepen injustice. Ethical design weighs the benefits of privacy against other core values like safety and fairness, always aiming for the least intrusive effective measure.

Summary#

We began with four simple but powerful ideas: do good, avoid harm, be fair, and use resources wisely. When we put them to work in real technology, they pull in opposite directions. Security can shut people out. The drive for perfect privacy can put people in danger. The worst burdens often fall on those with the least power. There is no formula to make these conflicts vanish, but we can stay curious, ask who is affected, and never pretend that a technical choice is just a technical choice. That mindset is the real heart of ethical technology.