Cheating Diesel Emissions Tests
The diesel emissions case investigates conflicts within professional codes of ethics when every available option seems to violate some ethical principle. The case is fictional but based on the Volkswagen emissions scandal in which software was used to detect when diesel vehicles were undergoing official emissions testing and reduce emissions during the test while allowing much higher emissions while driving. In this version of the case, an engineer suspects that colleagues have manipulated the software to mislead regulators, but the only way to verify the wrongdoing involves deception, bribery, or other misconduct. The central ethical question is what an engineer should do when one professional duty—disclosing factors that endanger the public or environment—appears to conflict with another duty, such as avoiding false, malicious, or harmful actions toward colleagues. This case highlights the limits of professional codes: they are useful ethical guides, but they do not always provide a simple answer when principles conflict.
Martin’s version
“The details of the following case are based on theVolkswagen emissions scandal that surfaced in 2015, but some pieces of information have been modified with the aim of demonstrating that internal conflicts in the IEEE code are possible.Imagine that you work for a US subsidiary of a large German automaker. Rumor hasit that engineers responsible for the software controlling the company’s diesel engines have manipulated the software. By using clever algorithms, the software detects if the caris being driven in regular traffic or is undergoing a test cycle of the type prescribed for the Environmental Protection Agency’s (EPA’s) emissions test. If the car senses that it is undergoing an emissions test, the software instructs the engine to reduce power, which reduces the emissions. However, if the software senses that the caris being driven on a normal road, the software instructs the engine to work as normal, with emissions that are far above those allowed by the EPA.
When you contact your colleagues working in the software department about this,they assure you that everything is okay. They have not manipulated the software to mislead the EPA, they say. However, you find this hard to believe because it is obvious that the emissions are higher than you would expect given the stellar performance in the official EPA emissions test. You conclude that your colleagues are probably lying.However, the only way to check this would be to perform some “false or maliciousaction.” You could, for instance, mislead your colleagues to believe (falsely) that your manager has authorized you to review the source code. Or, if we wish to make the example more extreme, we could imagine that you bribe a junior employee in the software department to email you copy of the source code, and in return for this you invite him to a conference in a fancy hotel (Figure 2.4).
Your method for verifying the truth of your hypothesis violates principle 9 of the IEEE code because you have, by “false or malicious action,” obtained information that is very likely to injure “others, their property, reputation, or employment.” However, per principle 1 of the IEEE code, you did nothing wrong because you have merely fulfilled your obligation to “disclose promptly factors that might endanger the public or the environment. The software in this diesel engine senses whether the car is undergoing an emissions test. If so, it instructs the engine to reduce power, which gives a misleadingly good test result. Is it morally wrong to obtain information about this form of cheating by some “false or malicious action” that is likely to injure “others, their property, reputation, or employment”?
The upshot is that no matter what you do, you will violate at least one principle of the IEEE code. The facts just happen to be such that you will violate either the first or the ninth principle of the IEEE code. There is no alternative action you could perform that would satisfy all principles of the code. What *should* you do?
Peterson, M. (2025). *Ethics for engineers* (2nd ed.). Oxford University Press.