Living in the age of networked dishonesty

For decades we've assumed that technology would make markets more efficient. It has. What we perhaps didn't anticipate is that it would also make theft, lies and deceit dramatically more efficient.

This is illustrated by three completely unrelated stories that caught my attention this week.

One examined how consumers are abusing chargeback rules through "friendly fraud". Another explored how candidates are using AI to cheat during job interviews. The third looked at how brands quietly fill their products with cheaper ingredients to protect margins.

At first glance, these stories appear to have nothing in common.

One concerns payments, another recruitment and the third consumer goods, yet I would argue they are all symptoms of the same phenomenon. They demonstrate how lies are beginning to overwhelm trust.

For decades, we have celebrated efficiency as one of the defining characteristics of modern business. We optimise supply chains, manufacturing, marketing, customer acquisition and logistics. Investors optimise returns. Consumers optimise prices. AI promises to optimise almost everything.

Efficiency and optimisation have become synonymous with good management.

The problem is that optimisation does not stop when it reaches human behaviour and, increasingly, people are no longer asking whether something is the right thing to do. They are asking whether it is possible to do it without suffering any consequences. That subtle shift changes everything.

The consumer who abuses the chargeback process knows the retailer will probably decide that fighting the claim costs more than simply accepting the loss. The candidate who uses AI during an interview knows they may be able to learn the job afterwards. The manufacturer who quietly substitutes cheaper ingredients hopes the change is small enough that most customers won’t notice or won’t care.

None of these decisions involve breaking the law. That is precisely why they are so interesting.

They represent rational behaviour within systems whose incentives have become slightly distorted.

Everyone is making individually logical decisions, yet collectively they erode something far more valuable than the money involved. They slowly consume the reservoir of trust on which every market depends.

This is not a new phenomenon. Banks have dealt with moral hazard for centuries. Insurance companies have always understood adverse selection. Economists have long recognised that incentives shape behaviour. What has changed is the speed, scale and sophistication with which optimisation can now occur.

This is why I often use the themes from Latin in my presentations: Caveat Emptor and Uberimma Fides, buyer beware and utmost good faith. These are two sides of every deal. Can I trust the seller to deliver? Can I trust the buyer is honest?

The issue we deal with now is that artificial intelligence has dramatically reduced the cost of pretending. It has become easier to present yourself as more knowledgeable than you really are and easier to automate customer interactions. More importantly, it has become remarkably easy to create convincing versions of reality itself.

Deepfake technology can now reproduce voices and faces with alarming accuracy. Criminals are already using AI to impersonate family members, colleagues and senior executives in attempts to persuade victims to transfer money.

Romance scams have become more sophisticated as AI generates believable conversations over weeks or months, creating emotional relationships that never really existed.

Authorised Push Payment (APP) fraud has become one of the fastest-growing forms of financial crime precisely because victims genuinely believe they are sending money to someone they know and trust.

The payment itself is authorised. The trust behind it is fabricated.

This explains why APP fraud has become such a difficult problem to solve.

Banks can build ever more secure payment systems, but they cannot easily distinguish between a genuine customer paying a genuine solicitor and a genuine customer who has been expertly manipulated into believing a criminal is their solicitor. The payment is technically correct. The trust behind it is false.

Why is this? Well, it’s worth digging back into the history of trust as, throughout history, trust has evolved.

First, we trusted people because we knew them. Then we trusted institutions because they could scale confidence beyond our immediate communities. Now, AI is introducing a third model, where trust increasingly depends on continuous verification rather than familiarity or institutional reputation alone.

Biometrics, behavioural analytics, cryptography and digital identity become the foundations on which trust is established. Verification is becoming the infrastructure that allows trust to scale.

But now there is a subtle and profound shift.

Traditionally, fraudsters stole your password, your card details or your identity. Today they are increasingly stealing something much more valuable. They are stealing your confidence.

Deepfakes, cloned voices and AI-generated conversations do not attack the security of the payment. They attack the judgement of the person making the payment. The objective is no longer to hack the system. It is to persuade the customer to bypass the system altogether.

Whenever the cost of deception falls, the cost of verification rises.

That is the hidden story behind these seemingly unrelated articles. Retailers invest more heavily in fraud detection because chargeback abuse increases. Employers redesign recruitment processes, because interviews no longer provide confidence that the candidate possesses the skills they appear to have. Food manufacturers spend millions analysing how far they can reformulate a product before customers lose confidence in it. Banks invest billions in behavioural analytics, confirmation-of-payee services, biometric authentication and AI-driven fraud detection because the traditional signals of trust have become easier to manipulate.

The result is what I think of as a trust tax.

Trust is one of the most efficient forms of economic infrastructure ever invented because it eliminates friction. When I trust my employer, my bank, my supermarket or the person on the other side of a transaction, I do not need endless verification, legal processes or compliance checks. The moment trust weakens, those costs begin to appear everywhere.

We already see the trust tax every day.

Banks ask us to confirm payments twice or sometimes more. By way of example, I recently bounced from retailer to PayPal to AMEX to bank just to make a simple payment for gaming on Roblox.

Apps demand biometric authentication. Retailers delay refunds while they investigate. Employers add extra interview rounds. Every additional step feels inconvenient, but every one exists because somebody, somewhere, found a way to optimise dishonesty.

This is why I increasingly believe the next decade of finance will be defined less by moving money and more by proving intent. A payment is easy. Knowing that the person making it understands what they are doing, has the authority to do it and has not been manipulated into doing it is becoming the real challenge. The transaction is no longer the difficult part. Establishing confidence in the transaction is.

It also explains why banking is so widely misunderstood.

We like to think banks move money, but payment systems move money remarkably well on their own. The unique role of banks has always been to make trust scalable. Credit assessment, fraud monitoring, identity verification, compliance and risk management are all mechanisms for answering a remarkably simple question: can this transaction, this customer or this organisation be trusted?

For thirty years, banks have competed to remove friction.

Faster payments, one-click commerce, instant onboarding and seamless digital experiences have all been celebrated as progress. AI turns that logic on its head. If deception becomes dramatically cheaper, friction stops being a failure of design and starts becoming a feature of trust. Asking a customer to pause before making a payment, confirming the beneficiary, checking a biometric or requesting a second approval are no longer signs of inefficient banking. They are evidence that trust has become too valuable to assume.

Much has been written about data being the new oil and AI becoming the defining competitive advantage of the coming decade. Both observations contain some truth, but they miss what may prove to be the scarcer resource as synthetic content, AI agents and automated decision-making are becoming commonplace. It means that organisations that can demonstrate genuine authenticity will become increasingly valuable (take note Meta).

The winners of the next decade will not simply be those with the smartest algorithms. They will be those capable of proving that their products are what they claim to be, that their employees possess the skills they demonstrate, that their AI systems act within agreed authority and that the person asking you to move £50,000 really is your son or daughter, your solicitor or your chief executive.

Perhaps that is the real lesson behind these three stories. We are entering an economy where almost everything can be generated, simulated or optimised. AI can write essays, conduct interviews, clone voices and create relationships that never existed.

In that world, authenticity becomes the rarest commodity of all.

For years, technology has been removing friction from commerce. In the future, we will use technology to put some of it back. It won’t be because we've forgotten how to build efficient systems, but because trust has become too valuable to assume.

That changes the role of banks fundamentally because, in an economy where almost everything can be generated, simulated or optimised, trust is no longer a by-product of finance.

It is the product.