The following information Was generated by asking ChatGPT to write an article on detecting and deterring theft and fraud in the stockroom and loading dock.

Interestingly, some of the information came almost word for word from my website, detecttheft.com, while I also discovered that two other sites had copied exact passages from my website and claimed it as there own.

So the question, then, is “how good is AI?”

Currently, a great deal of AI relies on gathering information across myriad sources and parsing it, generating content that has the greatest consensus across many sources. That doesn’t make it correct. It makes it commonly accepted, and there is a difference.

First, I’ll let you see what AI produced, then we’ll examine it at the end, both for what it does well and what it does poorly.

Introduction

Theft and fraud in the stockroom or loading dock can pose significant risks to Canadian businesses, resulting in financial losses, operational disruptions, and diminished employee morale. These areas, often bustling with activity and holding valuable inventory, are particularly vulnerable to internal and external threats. Understanding how to detect and deter theft and fraud is essential for safeguarding assets and ensuring business continuity.

Common Forms of Theft and Fraud in Stockrooms and Loading Docks

• Employee Theft: Staff may intentionally misappropriate goods, manipulate inventory records, or collude with external parties for personal gain.
• Vendor Fraud: Suppliers or delivery personnel might under-deliver, overcharge, or substitute inferior goods.
• External Theft: Unauthorised individuals may gain access to the premises, especially if security is lax.
• Inventory Manipulation: Falsifying records to hide losses, cover up theft, or create fraudulent stock discrepancies.

Detecting Theft and Fraud

Early detection is critical. The following strategies can help identify irregularities in the stockroom or loading dock:

Regular Inventory Audits:

1. Conduct frequent and random inventory counts. Discrepancies between physical counts and inventory records may signal theft or fraud.

Surveillance Systems:

• Install security cameras covering key areas such as entrances, exits, and storage racks. Modern systems allow for real-time monitoring and playback of footage to investigate suspicious activity.

Access Control Logs:

• Track who enters and exits the stockroom or loading dock. Electronic badge systems or ABM-controlled doors can provide detailed logs.

Transaction and Activity Monitoring:

• Review transaction histories for unusual patterns, such as repeated adjustments, voids, or returns. Automated software can flag anomalies.

Employee Behaviour Observation:

• Watch for changes in employee behaviour, such as reluctance to take holidays, secretive actions, or unexplained affluence.

Anonymous Reporting Channels:

• Encourage staff to report suspicious activities through anonymous tip lines or suggestion boxes.

Deterring Theft and Fraud

Prevention is often more effective than detection. Implement these measures to deter theft and fraud in your stockroom or loading dock:

• Strong Physical Security:
• Utilize secure locks, fencing, and alarm systems. Ensure loading docks are well-lit and access points are monitored.
• Clear Policies and Procedures:
• Develop and communicate clear inventory handling protocols. Educate staff on the consequences of theft and fraud.
• Segregation of Duties:
• Assign different employees to inventory receipt, record keeping, and auditing. This reduces the opportunity for collusion or cover-ups.
• Employee Screening:
• Conduct background checks during hiring and periodically review staff records for any red flags.
• Visible Surveillance:
• Prominently display security cameras and signage indicating monitoring. The perception of being watched can deter would-be offenders.
• Regular Training:
• Offer ongoing training on ethical behaviour, fraud prevention, and security protocols.
• Random Inspections:
• Carry out unannounced checks of inventory, delivery vehicles, and staff belongings.

Canadian Context and Legal Considerations

In Canada, employers must balance security measures with respect for privacy and labour laws. Surveillance and searches should comply with provincial legislation and employee rights. Establishing a respectful workplace culture, combined with robust security, can help prevent theft and fraud without alienating staff.

Conclusion

Theft and fraud in stockrooms and loading docks can be costly, but with vigilance, sound procedures, and a culture of accountability, businesses can significantly reduce their risks. By combining effective detection methods with strong deterrence strategies, Canadian companies can protect their assets and maintain trust among employees and stakeholders.

Ok. So how good is this article? First, the information is broadly accepted as valid. But these tips are fairly easy to understand simply by observation, and many of the sources are relying on practical experience.

But what happens if we ask the question slightly differently. I asked which methods cause the greatest loss, and AI cited collusion and falsifying records. But the new article ignored repeated thefts of small or medium items, failure to receive goods (cargo theft), mispricing and a host of other techniques. Collusion was barely mentioned in the original article.

This leads to an interpretation that AI is reactive, and not intuitive. It will not come up with ideas that are still emerging. So it’s a little like trying to stop a stampede from the rear. Until you get in front of the problem, you can’t direct it.

I conclude that AI is an effective tool for summarizing what we already know in this field. That is good enough, but not great. It is not an intuitive predictor, but rather a predictor based on history.

The second problem also turns on the intuition (or lack of it) that AI demonstrates. There are a host of issues in deterrence and detection that are not addressed.

So as part two of this podcast, I want to fill you in on several of these issues, so you can get in front of any problems you have in your business, rather than just trying to catch up.

Barcodes: Using replacement stick-on barcodes for SKU and inventory management allows employees, suppliers, drivers, etc. to substitute quantities, quality and even complete contents, since boxes rarely are opened and staff often are not well informed as to how to detect discrepancies. Do random inspections, use specific security tape to seal boxes, make incoming orders subject to approval and keep access to labellers and bar codes limited to select personnel.

Gate control. Gate control includes monitoring and controlling exist, stock to sales floor doors, office and internal egress and virtual gate controls. Physical restrictions to the stockroom are simple and easily monitored. However, virtual control requires electronic monitoring of access to documents, flow logs and financial data, so that an outside influence (other than designated personnel) cannot alter records. Because this is an emerging area, owners should be well-versed in trends and potential risks.

Rerouting. Rerouting of documents or merchandise is more discrete than direct theft, since there are intermediary steps involved that may cross lines of authority and responsibility, yet may initially seem innocuous. AI cannot predict these risks with a high degree of accuracy at this point.

Connecting lines of movement. Similar to, but the mirror image of rerouting, lines of movement, where there is a chain of events involved in stock or paperwork handling, allows for greater opportunity for deviance.

Repackaging. Repackaging may involve multiple items in a sealed package, incorrect items, or lower quality items. While cameras may be able to identify repackaging in the stockroom, those boxes may be manipulated outside of the business location, enroute, on the sales floor, and so on. Random checks will reduce opportunity.

Mules. Using innocent and unaware carriers to move your goods is an effective and not easily detected system, wherein the perpetrator includes items in an order being delivered outside of the business (home delivery, cross-docking, store-to-store, etc.). It can take many forms and makes assignment of blame difficult, since the oblivious mule is not accountable. However, mules often are willing accomplices.

Rearranging & staging. Staging and rearranging is a well-worn path for theft, but also for fraud, where a person moves goods from one location to a more private location in preparation for theft, or creates a work order, then an invoice to stage for a fraudulent transaction.

Working with office staff who have control of ordering & payments. AI recognizes the high risk of collusion, but, because computer systems, inventory management programs, sales systems, payroll issues and so on are becoming more automated, the ability to keep in front of emerging problems often is a challenge.

Separation of authority. Where lines of authority, communication and responsibility overlap, there are greater opportunities for manipulation.

Vigilance with people entering stockroom, including sale persons, customers, trades and repair persons, delivery, shippers, regular staff, cleaning staff is essential to control virtual and physical manipulation of inventory, money and data.

Reticketing. Reticketing is a strategy that has been employed by staff for a long time, but while physical ticket use has almost disappeared, ticketing and pricing on software systems allows for both inadvertent and intentional manipulation.

Duplicate paperwork. Duplicate paperwork, or using  pre-completed zip sets that allow for the top copy to be changed after authorization allow money and paperwork handlers (and, in some cases, stockroom personnel) to change orders, prices, or payments.

Businesses generally need to improve scan systems that automatically order and process data, so that they cannot be manipulated. Manual cross-checks are the most effective way to do so, including periodic forensic bookkeeping.

Paying on order or work order, rather than on invoice. Smaller operations sometimes pay immediately for drop shipments or “deal” purchases. Similarly, workorders for repair or maintenance may be paid directly to the tradesperson, then may be lured into paying an “outstanding” invoice at a later date.

Allowing more than one person in stock room to deliver or pick up opens the doors, literally, to being distracted by the chaos and having one of the multiple persons stealing or defrauding the receiver.

These are only a few of the techniques whereby fraud or theft occurs, most of which are not easily determined by AI, whether that AI is simply a query about risk and probabilities, or AI-generated recognition software.

AI is a support mechanism, particularly at this early period in adoption.

As A final test of the ability of AI to generate predictive responses, I asked the following:

My stockroom has one receiving door, two entrances to sales floor, one entrance to office, one exit to staff parking lot. We sell mostly produce. i have no specific shipper receiver. what are the most likely ways that theft or fraud will happen?”

AI responded this way:

“In a stockroom setting with multiple access points, no dedicated shipper/receiver, and high-turnover, perishable goods like produce, theft and fraud are most likely to occur through internal employee schemes, receiving discrepancies, and collusion with external parties.”

It went on to cite grazing & concealment, sweethearting, return fraud, theft of cash, time theft and unauthorized access as the most common internal theft & fraud problems.

For receiving and inventory vulnerabilities, it suggested that vendor fraud or collusion, fraudulent receiving errors and inventory errors were the most common.

And for external theft is declared that smash and grab and shoplifting from sales floors were the biggest problems.

These are old-world solutions that, while still valid concerns, do not consider emerging problems, such as the ones discussed above and the dozens of others not included in this article.

The conclusion: Use AI only as one of several tools at your disposal. Recognize that the best way to understand, predict, anticipate and respond to human behaviours is to use humans to do so.

AI can be a tool, but it is simply not the solution.