The Software That Knew When to Behave
The elegant simplicity of the fraud was what made it so devastating. When a Volkswagen diesel engine detected the telltale signs of laboratory testing—stationary wheels, controlled temperature, predictable throttle patterns—software buried deep in its electronic control unit would shift the vehicle into a special mode. Emissions controls that remained dormant during normal driving would spring to life. Nitrogen oxide output would plummet to legal limits. The test would pass. Then, back on actual roads with actual drivers, the software would return the engine to its default state: powerful, efficient, and releasing nitrogen oxides at levels up to 40 times the legal threshold .
Volkswagen admitted to fitting this computer software to cheat emissions data during tests . The software, legally defined as a 'defeat device' prohibited under the Clean Air Act , represented more than a technical violation. It was a deliberate architecture of deception, engineered not by rogue programmers but embedded in the product development process of what was, at the time, the world's second-largest automotive manufacturer.
That was 2015. What has emerged in the years since suggests that Volkswagen's confession was not the ending of a contained corporate scandal but the opening of a far more disturbing chapter: evidence that defeat devices, software manipulation, and systematic emissions cheating were not aberrations but industry practice, woven into the diesel supply chain from Detroit to Stuttgart to Tokyo.
The Widening Spiral
Cummins, the Indiana-based engine manufacturer whose logo appears on Ram pickup trucks across America, will pay a $1.675 billion penalty for diesel engines fitted with 'defeat devices' that allowed excessive nitrogen oxide levels in everyday use . The settlement, announced in 2023, covers 600,000 vehicles—a scale that dwarfs many earlier cases and positions Cummins's fine as one of the largest environmental penalties in United States history.
The technology in question mirrors Volkswagen's approach: software calibrations that recognised test conditions and modified engine behaviour accordingly. What differs is the vintage. Where Volkswagen's fraud became public in 2015, Cummins's defeat devices were installed in vehicles manufactured between 2013 and 2019—well after the scandal's initial exposure, well after the industry claimed to have learnt its lesson.
The same pattern appears in Federal Motor Transport Authority records. In 2018, Germany's KBA ordered BMW to recall over 11,000 cars due to inadmissible defeat devices in diesel engines . That same year, the KBA imposed a mandatory recall on Daimler AG for an illegal shutdown mechanism in the GLK 220 diesel model . US investigation files, according to German reporting, heavily implicate Daimler in parallel emissions violations .
Renault, facing pressure after the Volkswagen revelations, recalled more than 15,000 new cars to check and adjust their filtration systems . Yet independent testing revealed the depth of the problem: Renault's Espace Energy dCi 160 emitted over ten times more nitrogen oxides in realistic driving tests than in the EU's laboratory-based regulatory protocol . The divergence was not marginal; it was categorical.
"When software knows the difference between a test and reality, and actively exploits that difference, we are beyond engineering optimisation. We are in the territory of fraud."
Fiat's diesel engines exceeded the Euro 6 pollution threshold by an average of 18 times, with the 500X model reaching nitrogen oxide emissions of 0.080 grammes per kilometre—a figure that makes a mockery of the regulatory ceiling . In South Korea, regulators moved to fine Nissan for manipulating emissions data on a popular diesel SUV . Even Toyota, a manufacturer whose hybrid credentials have insulated it from much diesel scrutiny, now faces a class-action lawsuit in Australia over claims that up to 500,000 diesel-powered vehicles contained emissions defeat devices . Toyota Australia lost its Federal Court appeal in 2022 regarding diesel particulate filter failures in HiLux, Fortuner, and Prado models , a ruling that leaves the company exposed to substantial damages claims from affected owners.
The Belgian Raids
In early 2016, inspectors from Belgium's Economic Inspectorate arrived unannounced at Opel dealerships across the country . The raids followed reports that dealers were secretly adjusting the software of polluting Opel Zafira vehicles —allegedly at the direction of Opel Belgium itself .
The allegations were precise: technicians were installing software updates ostensibly to address emissions concerns, but the modifications were occurring outside official recall channels and without the transparency that regulatory frameworks demand. Febiac, the Belgian automotive trade association, publicly supported Opel's denials, stating there was 'no explanation for the results' that testing had produced . Yet the Economic Inspectorate's intervention suggested authorities found the evidence compelling enough to warrant criminal investigation.
What made the Opel case particularly resonant was its suggestion that emissions fraud had metastasised beyond manufacturers into the dealer network—that the pressure to meet regulatory standards whilst maintaining vehicle performance had created incentives for manipulation at every level of the supply chain.
The American Exception
When the Environmental Protection Agency accused Fiat Chrysler in 2017 of installing eight undisclosed pieces of software that could alter how vehicles emitted air pollution , the company's response differed sharply from Volkswagen's. Fiat Chrysler insisted the software was necessary for engine protection and durability—that it was not a defeat device in the legal sense but a legitimate engineering solution that happened to affect emissions.
The distinction was important. A defeat device, by definition, serves no purpose except to circumvent emissions standards. Software that protects engine components from damage, even if it incidentally increases emissions, occupies murkier legal territory. The EPA's position was that any software reducing emissions controls during normal driving constitutes a defeat device regardless of its secondary engineering justifications. Fiat Chrysler's position was that this conflated fraud with design trade-offs inherent to diesel engineering.
The debate was not merely semantic. It went to the heart of whether the diesel emissions scandal represented criminal conspiracy or the collision between regulatory idealism and thermodynamic reality. Internal combustion engines operate through controlled explosions at temperatures exceeding 1,000 degrees Celsius. Emissions control systems—exhaust gas recirculation, selective catalytic reduction, diesel particulate filters—impose constraints that can, under certain conditions, damage engine components or reduce performance to levels consumers and fleet operators would find unacceptable.
The counter-argument, advanced by regulators and environmental groups, was that these engineering challenges were known when manufacturers lobbied for diesel-friendly emissions standards. The standards themselves represented negotiated compromises between environmental protection and technical feasibility. To then claim those standards were unrealistic was to admit participation in a regulatory charade.
The Precedent That Wasn't
On 22 October 1998, the Environmental Protection Agency announced an $83.4 million penalty against diesel manufacturers—the largest Clean Air Act settlement in history at that time . The case involved seven engine manufacturers, including Mack Trucks, who had installed software that increased fuel economy during highway driving by reducing emissions controls. The parallel to the Volkswagen scandal, 17 years avant la lettre, was unmistakable.
Yet the 1998 settlement failed to function as deterrent or warning. The penalties, whilst record-breaking, were absorbed without fundamentally altering industry practice. The legal frameworks remained unchanged. The detection methodologies that allowed manufacturers to identify test conditions and modify engine behaviour remained in place. The regulatory cycle—a new emissions standard announced, a compliance deadline set years in the future, manufacturers' assurances that technical solutions were forthcoming—continued uninterrupted.
What the 1998 case revealed, in retrospect, was not merely that defeat devices existed but that the regulatory system was structurally incapable of preventing them. Laboratory testing protocols were, by necessity, standardised and therefore predictable. Any standardised test creates an optimisation target. Once the target is known, engineering resources will be deployed to meet it—through genuine emissions reduction if possible, through creative compliance if necessary, through outright fraud if the risk-reward calculation permits.
The Detection Gap
The German Federal Motor Transport Office has not initiated legal proceedings against Porsche and Audi for their fraudulent exhaust cleaning systems for more than three years, despite evidence of violations . The administrative paralysis reflects a deeper problem: detection and enforcement capabilities have not kept pace with the sophistication of engine control software.
Modern automotive software contains millions of lines of code, much of it proprietary and protected as trade secrets. Regulators lack the resources and often the technical expertise to audit this code comprehensively. Laboratory testing, even when supplemented by on-road measurements, captures vehicle behaviour only under the specific conditions of the test. Software can be calibrated to recognise those conditions with extraordinary precision—monitoring not just wheel speed and steering angle but ambient temperature, barometric pressure, even the duration since the engine was started.
Independent testing has exposed the scale of real-world deviation. Data from realistic driving tests showed that diesel engines emitted far more pollution than European regulatory limits , with nitrogen oxide levels routinely exceeding laboratory results by factors of five, ten, even twenty . But translating these discrepancies into legal cases requires proving intent—that software deliberately targeted test conditions rather than incidentally behaving differently under varied operating parameters.
The Regulatory Compromise
What has emerged from a decade of scandal and litigation is an uncomfortable recognition: the regulatory framework for diesel emissions was premised on assumptions that proved false. Manufacturers promised that advanced emissions control technology could deliver both environmental compliance and the performance characteristics—torque, fuel economy, durability—that made diesel attractive. Regulators structured standards around these promises. Consumers and fleet operators made purchasing decisions based on advertised specifications.
The defeat device revelations suggest this was a trilateral failure. Manufacturers made commitments they could not or would not honour. Regulators accepted assurances without adequate verification mechanisms. And the market rewarded claimed efficiency whilst remaining wilfully indifferent to the methods by which it was achieved.
The economic logic was perverse but powerful. A manufacturer that genuinely complied with emissions standards whilst competitors deployed defeat devices would produce vehicles that were less powerful, less efficient, and more expensive—a market disadvantage so severe it threatened viability. The incentive structure rewarded cheating and penalised honesty, a dynamic that explains why the scandal sprawled across nearly every major manufacturer rather than remaining confined to a single bad actor.
The Unresolved Question
A decade after Volkswagen's admission, the fundamental question remains unresolved: can diesel technology meet contemporary emissions standards under real-world driving conditions at a cost consumers will accept? The technological optimists insist the answer is yes—that advanced selective catalytic reduction systems, combined with improved engine calibration and better-quality diesel fuel, can achieve genuine compliance. The sceptics point to the evidence: test after test showing persistent, substantial divergence between laboratory and real-world emissions, even in vehicles produced years after the scandal broke.
The market has begun rendering its own verdict. Diesel's share of new car sales in Europe has collapsed from 55 per cent in 2011 to less than 20 per cent by 2022. Manufacturers have shifted investment toward petrol-electric hybrids and battery-electric vehicles, technologies that avoid the diesel dilemma entirely. Cities across Europe have implemented or are planning diesel bans in urban centres. The fuel that was marketed as the environmentally responsible choice—lower carbon dioxide emissions than petrol, thanks to greater efficiency—has been recast as a public health threat, implicated in tens of thousands of premature deaths from respiratory and cardiovascular disease.
Yet millions of diesel vehicles remain in operation, many of them purchased in good faith by consumers who believed manufacturers' environmental claims. The recall programmes and software updates that have followed the scandal's exposure address only some vehicles and, in many cases, reduce performance or fuel economy—imposing on individual owners the costs of manufacturers' fraud.
The Question of Accountability
Cummins's $1.675 billion penalty is substantial, but the company's 2022 revenue exceeded $28 billion. Volkswagen's total costs from the diesel scandal—fines, recalls, legal settlements, vehicle buybacks—have exceeded $30 billion, yet the company remains profitable and its senior leadership largely unchanged. BMW, Daimler, and Renault have paid penalties and implemented recalls whilst maintaining that their actions fell short of fraud, characterising them instead as overly aggressive optimisation or misunderstandings of regulatory requirements.
Criminal prosecutions have been limited and largely confined to mid-level engineers and managers—individuals who implemented defeat device strategies but rarely designed them. The systemic nature of the fraud, the fact that it appeared across multiple manufacturers, divisions, and continents, suggests decision-making at senior executive and board levels. Yet corporate structures, legal strategies, and the difficulty of proving individual knowledge have largely insulated those levels from criminal liability.
Environmental groups and public health advocates argue that the penalties imposed, whilst numerically large, remain insufficient to alter corporate behaviour fundamentally. The cost of genuine compliance—redesigning engines, accepting performance trade-offs, potentially ceding market share—may still exceed the expected cost of fraud when the probability of detection and the scale of likely penalties are factored in.
What Was Lost
Beyond the legal and financial reckonings, the diesel emissions scandal has corroded something less quantifiable but perhaps more significant: institutional trust. The automotive industry enjoyed substantial regulatory latitude, predicated on the assumption that manufacturers' technical expertise and reputational concerns would ensure responsible self-governance. That latitude has been revoked. The scandal has empowered regulators to demand real-world emissions testing, to require greater transparency in engine control software, and to treat manufacturers' compliance claims with scepticism rather than deference.
But the damage extends beyond the industry itself. The scandal has become a template for understanding how complex technical systems can be manipulated by sophisticated actors, how regulatory frameworks can be gamed, how corporate assurances can be systematically false. It has fed a broader erosion of confidence in institutions—governmental, corporate, scientific—to act in the public interest when commercial pressures point elsewhere.
The diesel defeat device was, in the end, a machine that knew when it was being watched and altered its behaviour accordingly. That this was engineered not by a single bad actor but embedded across a global industry suggests a failure more profound than fraud. It suggests an industrial culture that had lost sight of the distinction between engineering cleverness and ethical responsibility, between optimising for a test and solving the problem the test was meant to measure.
The vehicles that carried these devices are still on the roads, still emitting nitrogen oxides at levels that sicken and kill. The companies that built them are still in operation, still manufacturing, still advertising their environmental credentials. The regulatory frameworks that failed to prevent the fraud have been modified but not transformed. And the question that haunts the entire saga remains unanswered: if defeat devices were this widespread, this persistent, this systematic, what else have we been systematically lied to about?