Workplace fraud may be becoming increasingly normalised, with 87% of finance professionals admitting they have ignored a claim they believed was fraudulent and 67% saying they could submit a dishonest claim themselves if colleagues were doing the same. The Medius Financial Census 2026 also reveals widening gaps between AI adoption, governance, employee training and wellbeing.
Small acts of financial dishonesty are becoming increasingly tolerated in workplaces, raising questions about corporate culture, employee trust and whether organisations are focusing on sophisticated external fraud while overlooking misconduct happening inside their own businesses.
That’s one of the headline findings from the new Medius Financial Census 2026, based on research involving 2,386 finance executives across the US, UK, Sweden and France. Some 87% of finance professionals surveyed said they have ignored a small expense, reimbursement or claim they believed was fraudulent. Even more concerning, 67% said they would be likely to engage in a minor dishonest expense claim themselves if the behaviour were common among colleagues.
Nearly three-quarters (74%) believe small forms of fraud resulting in relatively minor financial losses are already common in workplaces, while 64% said they could feel justified committing a minor dishonest financial act if they felt underpaid or undervalued. More than half (57%) also admitted they would round up an expense or mileage claim if they believed it would go unnoticed.
KEY FINDINGS
Key findings from the report show that:
- 87% of finance professionals have ignored a small expense, reimbursement or claim they believed was fraudulent.
- 67% could submit a dishonest claim themselves if that behaviour was common among colleagues.
- 64% could justify a small dishonest financial act if they felt underpaid or undervalued.
- 57% would round up an expense or mileage claim if they thought nobody would notice.
- 93% are concerned about AI-generated fraud during the next 12 months.
- 38% already have agentic AI operating within at least some finance processes, with another 50% planning deployment within 12 months.
- 86% say employees’ AI use now factors into performance evaluations.
- 75% say AI usage has increased worker fatigue or burnout.
- Only 27% report receiving comprehensive AI training through structured programmes with dedicated time.
The Medius research suggests finance functions are reaching an uncomfortable inflection point: automation and AI adoption are accelerating, but fraud, governance gaps, late payments and pressures on employees remain.
WORKPLACE FRAUD BECOMING NORMALISED
Medius describes the low-value financial rule-breaking highlighted in its research as “shallowfakes” – small acts such as embellishing mileage or expenses that can appear insignificant individually but potentially accumulate into considerable losses when repeated across an organisation.
It is important to distinguish the company’s use of the term from “deepfake” fraud involving AI-generated impersonation. Here, the problem is often much more mundane: adding a few miles to an expense claim, submitting a receipt twice, inflating a reimbursement or using a company subscription for personal purposes.
Medius argues that the real warning sign is not simply whether employees recognise that behaviour as questionable. It is whether they have begun to accept it. When two-thirds of finance professionals surveyed say they could participate in minor financial dishonesty if colleagues commonly behaved in the same way, ethical boundaries risk becoming socially reinforced. And when 87% say they have already ignored something they believed was fraudulent, the issue potentially moves beyond fraud controls into workplace culture.
SHALLOWFAKE FRAUD
Chris Wilmot, Chief Financial Officer at Medius, warned that organisations may be concentrating too heavily on high-profile sophisticated scams while smaller financial losses accumulate. “Costly deepfake fraud gets all the headlines,” said Wilmot. “But while finance professionals keep a sharp eye out for these scams, hundreds of thousands of dollars are slipping out the back door through shallowfake fraud.
These “seemingly minor ‘micro frauds’ add up to death by a thousand cuts for organisations without the controls in place to catch them”, adds Wilmot. That makes an organisation’s response to seemingly minor transgressions significant.
Employees observe not only formal rules, but what happens when those rules are broken. If questionable claims routinely go unchallenged, workers can receive a powerful informal message about what their organisation really considers acceptable.
AI IS ALREADY BEING USED TO FAKE EXPENSES
The findings come as separate research suggests artificial intelligence is already making some forms of expense fraud easier to carry out. As Fair Play Talks reported in July, research commissioned by Emburse found that 40% of US employees and 29% of UK workers surveyed said they had used AI to create or manipulate expense receipts.
Among US workers surveyed, 19% said they had used AI to completely fabricate a purchase, while 15% had increased the value of an existing purchase. Some 40% of those generating fake receipts said they had used company-funded AI tools to do so.
The study also revealed an important financial-pressure dimension. Some 27% of US respondents admitted passing personal purchases off as business expenses because of financial pressures.
Together, the Emburse and Medius findings suggest organisations potentially face fraud from two directions. New technology is making falsification easier, while workplace attitudes may make relatively small acts of dishonesty easier to rationalise.
DOES FEELING UNDERVALUED CHANGE ETHICAL BEHAVIOUR?
One of the most interesting Medius findings from an employee perspective is that 64% said they could feel justified committing a small dishonest financial act if they felt underpaid or undervalued. That does not establish that feeling poorly paid or undervalued causes fraud. Nor does dissatisfaction excuse deliberate dishonesty.
But the finding raises an important question about how employees rationalise behaviour they know breaks workplace rules. If workers begin regarding an inflated expense as recompense for inadequate pay, excessive workloads, delayed reimbursements or feeling unappreciated, a compliance issue can become entangled with wider perceptions of organisational fairness.
For employers, that makes the conversation about more than fraud detection. It also touches on fair pay, employee experience, organisational trust, leadership and culture.
SMALL TRANSGRESSIONS TO ETHICAL BLINDNESS
The findings also connect with a longer-running debate about why otherwise ordinary employees participate in, tolerate or remain silent about questionable behaviour. Fair Play Talks previously explored the concept of “ethical blindness”, highlighted by Professor Guido Palazzo, Professor of Business Ethics at the University of Lausanne.
Palazzo has argued that corporate scandals cannot always be explained simply by identifying individual “bad apples”. Organisational environments, pressure, leadership signals and an absence of challenge can gradually cause inappropriate practices to become normalised.
That context makes the Medius finding that 87% of finance professionals surveyed have ignored suspected minor fraud especially striking. The critical issue is not only whether someone submits a questionable claim. It is what everyone around them does when they see it.
WHY ARE EMPLOYEES LOOKING THE OTHER WAY?
There may, however, be an important difference between employees who ignore behaviour because they regard it as harmless and those who stay silent because they fear the consequences of challenging it.
Previous Fair Play Talks coverage found that one-third of US employees surveyed said fear of retaliation or other negative consequences could prevent them from reporting unethical or illegal conduct. The Trust @Work study also found 22% had witnessed unethical or illegal conduct at work, while 21% had felt pressure to compromise their own ethical standards.
The comparison raises an important question for organisations confronted by the Medius figure. Why are management failing to intervene? Is it because they think a few extra miles on an expense claim do not matter? Do they believe everyone else behaves similarly? Do they assume the leadership team will do nothing? Or do employees fear being labelled difficult, disloyal or troublesome if they report a colleague? An organisation could potentially be dealing with several of those problems simultaneously.
AI FRAUD FEARS ARE SURGING
At the same time as finance professionals appear increasingly tolerant of some low-level internal fraud, their anxiety around sophisticated AI-enabled fraud is escalating. Some 93% of respondents said they are concerned about AI-generated fraud over the next 12 months.
US organisations surveyed reported average yearly losses of $168,000 due to invoice fraud, with respondents experiencing an average of one successful invoice fraud attempt each month, according to Medius. Generative AI can make fabricated invoices, receipts, payment requests and communications increasingly convincing, potentially making the distinction between genuine and fraudulent transactions more difficult.
But focusing purely on external technological threats risks overlooking something much closer to home. The Medius findings suggest businesses need to address both the technology that makes fraud easier and the culture that can make dishonesty acceptable.
AI ADOPTION IS OUTPACING GOVERNANCE
Finance functions themselves are rapidly adopting AI. Some 38% of finance executives surveyed already have agentic AI operating within at least some finance processes, while another 50% plan to deploy it within the next 12 months.
Yet governance appears to be lagging. Some 90% of finance professionals said there is always a financial or compliance threshold at which human approval is required, regardless of an AI system’s accuracy or track record. However, 46% said those thresholds are understood but not formally documented. That distinction matters.
An informal expectation that “a human will check” is very different from an auditable governance framework setting out exactly when intervention is required, who owns the decision and where ultimate accountability sits.
WHO IS RESPONSIBLE WHEN AI GETS IT WRONG?
The accountability gap becomes even more evident when organisations consider what happens after an AI mistake. When respondents were asked who would be responsible if an AI error caused financial loss or a compliance problem, answers were distributed almost evenly between IT leaders, finance leaders and employees acting on the AI recommendations.
Meanwhile, 45% of finance leaders said their teams often act on AI-generated recommendations without human intervention. Ahmed Fessi, Chief Transformation & Information Officer at Medius, said that gap will become increasingly significant as finance functions automate more decision-making.
“Organisations are growing comfortable allowing AI to influence decisions, yet many still lack clear ownership when those decisions go wrong,” said Fessi. “As autonomous finance becomes more common, explainability, governance, and accountability are just as important as accuracy.”
The findings highlight a crucial responsible-AI principle. Human oversight means little if nobody knows which human is actually accountable.
EMPLOYEES JUDGED ON AI BEFORE THEY ARE PROPERLY TRAINED
AI is not only changing finance processes. It is increasingly influencing finance careers. More than half (55%) of finance leaders surveyed said AI fluency has become a meaningful differentiator when recruiting employees.
Even more strikingly, 86% said employees’ AI use now factors into performance evaluations. Yet only 27% report receiving comprehensive AI training through structured programmes with dedicated time.
That exposes a potential workplace fairness problem. Organisations are increasingly expecting workers to demonstrate AI capability – and in some cases judging their performance on it – while many employees apparently remain without comprehensive structured training.
That risks creating a new divide between workers who already have confidence, access and experience with AI and those who have had fewer opportunities to acquire those skills. If AI fluency becomes “career currency”, employers also have a responsibility to ensure employees have a fair opportunity to earn it.
THREE-QUARTERS SAY AI IS INCREASING BURNOUT
There is another important contradiction. Artificial intelligence is frequently promoted as a way to automate repetitive tasks, reduce administrative workloads and allow employees to concentrate on more valuable work. But 75% of finance professionals surveyed said AI usage has increased worker fatigue or burnout.
The research does not demonstrate that AI itself directly causes burnout. But rapid technological change can bring additional learning demands, changing expectations, uncertainty about jobs and pressure to increase output while employees continue doing their existing work.
That means employers should look beyond adoption numbers. The real test is not simply how many employees use AI. It is whether AI is actually making work better.
AUTOMATION HASN’T SOLVED LATE PAYMENTS
The same principle applies to more established forms of finance automation. Some 85% of finance teams surveyed report using some level of accounts payable automation. Yet manual work and payment delays persist.
Medius found that 45% of organisations said between 21% and 40% of their invoices are paid late in a typical month, while 96% said managing late payments contributes to stress or burnout among AP teams.
The research points to internal approval bottlenecks, inadequate cash-flow management, disputed invoices or data mismatches, understaffing and poor visibility as key causes. The commercial consequences extend beyond finance teams. According to Medius:
- 45% said suppliers had imposed stricter upfront payment terms.
- 43% reported suppliers ending client relationships.
- 42% said suppliers had reduced service quality or speed.
- 42% reported disputes escalating to formal dispute or legal action.
Kevin Permenter, Research Director of Financial Applications at IDC, said technology investment alone is therefore not enough. “Automation was supposed to make late payments the exception, not the norm. Yet the findings suggest many organizations continue to experience payment delays despite ongoing technology investment. As supplier terms tighten and business disruptions grow, the challenge is no longer simply adopting automation, but ensuring those investments deliver measurable operational outcomes.”
WHAT EMPLOYERS SHOULD DO NEXT
The findings point to a challenge that stretches well beyond checking receipts more carefully. Here are a few pointers for employers:
Don’t dismiss small ethical breaches
Minor financial losses can seem inconsequential individually. But repeatedly overlooking behaviour because each incident is small can signal that certain rules are optional. Policies need to be understood, proportionate and consistently applied, including when the individual involved is senior, high-performing or commercially valuable.
Find out why employees and managers are looking the other way
An 87% non-intervention figure should prompt organisations to examine what sits behind employee silence. Employers should assess whether people view minor fraud as acceptable, believe reporting will achieve nothing or fear personal consequences for raising a concern. Those are different cultural problems and require different interventions.
Build a culture where employees feel safe to speak up
A whistleblowing hotline alone does not create psychological safety. Employees need confidence that reports will be assessed independently, concerns will be taken seriously and retaliation will not harm their careers. Leaders also need to demonstrate visibly that ethical standards apply consistently.
Examine fairness as well as fraud
The finding that 64% could justify minor dishonesty when feeling underpaid or undervalued should not become an excuse for fraud. But it does warrant attention. Pay practices, reimbursement speed, workloads, recognition and perceptions of fairness can influence trust between employees and organisations. Fraud controls should therefore sit within a broader culture of fairness and accountability.
Formally define AI accountability
Before giving autonomous systems more authority, organisations should establish clearly:
- what AI is permitted to decide;
- where mandatory human approval begins;
- who is responsible for reviewing recommendations;
- how decisions are documented and audited;
- how employees can challenge AI-generated outputs;
- and who ultimately owns an AI-related financial or compliance failure.
Responsible AI requires more than having a human somewhere “in the loop”.
Train people before judging them on AI
If AI fluency influences hiring, performance and career progression, organisations should provide meaningful opportunities for people to acquire those skills. That means structured learning, responsible-use guidance and protected time to develop competence – rather than simply introducing AI tools and expecting employees to teach themselves.
Monitor AI’s impact on wellbeing
AI implementation should include employee-impact measures alongside productivity measures. If new tools increase workload, learning pressure or expectations rather than reducing them, organisations need to understand why. Burnout should not be treated as an inevitable price of digital transformation.
MEASURE OUTCOMES, NOT TECHNOLOGY ADOPTION
Measure outcomes, not tech adoption
The late-payment findings provide a useful warning. Having automated a process does not mean the process is working well. Businesses should assess whether investment actually reduces manual intervention, errors, delayed payments and employee workload – and whether suppliers and workers experience the improvement too.
RESPONSIBLE BUSINESS & UNCOMFORTABLE CONTRADICTIONS
The Medius findings expose several uncomfortable contradictions inside modern organisations. Businesses are investing heavily in increasingly sophisticated technology to detect fraud, automate financial processes and introduce AI.
Yet some of their most significant risks remain profoundly human. When employees regard minor dishonesty as normal, colleagues look the other way, workers are frightened to report misconduct or people begin rationalising rule-breaking because they feel unfairly treated, technology alone cannot create an ethical organisation.
Nor should AI adoption automatically be equated with progress. When 86% say AI use affects performance evaluations, only 27% report comprehensive structured AI training and 75% say AI is increasing fatigue or burnout, employers need to consider whether expectations are advancing faster than the support available to workers.
The common thread running through fraud, AI governance, late payments and employee wellbeing is therefore culture and accountability. Controls matter. Technology matters. Policies matter. But so do the everyday signals organisations send about what behaviour is acceptable, whether speaking up is safe, who takes responsibility when things go wrong and whether technological change is genuinely improving work for the people expected to use it.
Major ethical failures do not always begin with a dramatic act. Sometimes they begin with something small that everyone can see – and nobody challenges.
Click here to download the Medius Financial Census 2026
RELATED ARTICLES
40% of US employees and 29% of UK workers have used AI to fake expense receipts.
Can a holistic approach to ethics and compliance minimise corporate scandals?
Workplace ethics crisis – one-third of employees fear retaliation for reporting misconduct.
Nearly seven in 10 employees now fear AI will lead to layoffs at their company within the next three years, according to a new workplace mental health report that paints a stark picture of rising burnout, worsening anxiety and collapsing trust in employers.
Sign up for our newsletter
