A 46-page academic paper by Richard Thompson Ainsworth of Boston University School of Law describes "zappers" - programs designed to divert some sales transactions from the normal sales processing and accounting systems. Fraudsters with sufficient access to an organization's sales systems (e.g. small business owners) sometimes use zappers either to misappopriate the entire sales income for the diverted sales (steal the entire value from the company - the sales don't go through the books) or to to manipulate the value (for example to steal the VAT/GST/sales tax content).
So-called "zap" and "super-zap" programs have existed for decades in the mainframe world. They allow intervention on databases, overriding normal access constraints to manipulate the data, and potentially programs, directly. They are supposed to be used only under carefully controlled emergency conditions, for instance to modify or delete a rogue data record that is somehow blocking an entire batch from processing. Most competent sysprogs (systems programmers) or systems administrators have the knowledge and capability to run zap programs and can potentially meddle with the systems in a virtually unstoppable and undetecable manner, if they are careful anyway: well-written programs have built-in integrity checks and other controls that at least identify and flag direct interventions. Unfortunately, if the sysprogs also have the capability to suspend or edit the audit trails, or substitute hacked programs, or subvert the operating system calls, or ... or ... all bets are off. Remember this possibility if you ever hear a sysprog for a financial institution bragging about the speed of his new Ferrari.
Going back to sales zappers, the article points out differences in the ways such frauds are detected in the UK and EU. In the States, it seems the evidence suggests that income tax investigations "often" (or rather occasionally!) catch zapper users, while in EU they are more likely to be caught by sales tax investigations. This begs the question: why not do both? And while you're at it, why not take a close look at those "shrinkage" stock losses - the ones that conceal employee as well as customer thefts of goods?