iComplai Predicts Food-Fraud Adulterations in Betel Nuts and Orange Juice
In June 2025 the models named two commodities, two tactics and two analytes. Between July and December, notifications naming those analytes rose 605% and 783%.
In June 2025, iComplai's early-warning models signalled a food fraud risk spike in betel nuts and orange juice, identified the likely tactic in each - artificial sweeteners in the nuts, artificial dyes in the juice - and named the analyte to test for. Between July and December 2025 the incident curves followed: notifications involving cyclamate in betel nuts rose 605%, and notifications involving Sunset Yellow FCF in orange juice rose 783%.
Why these two commodities?
Because both were under the same kind of pressure at the same time. Betel nuts and orange juice each faced margin compression and volatile supply, and those are the conditions that historically raise the reward for dilution, masking or colour adjustment. Fraud is an economic response before it is an analytical problem, which is why the commodity list can be drawn up before any incident exists.
The known tactics differ by product. In betel nuts: blending low-grade or spoiled nuts, boosting moisture, and masking the result with artificial sweeteners or flavours. In orange juice: dilution, blending in cheaper citrus juices, and adjusting colour with dyes. Both are textbook entries in a food fraud vulnerability assessment, which is where they belong long before a notification arrives.
What did the model flag, and on what?
Two separate risk drivers, one conclusion each.
- Betel nuts. The risk index peaked sharply on harvest loss and market tightness. The model raised an increased food fraud risk alert and surfaced artificial sweeteners as the high-probability tactic to test for.
- Orange juice. Heightened fraud risk driven by high prices, disease pressure and extreme weather. The model highlighted artificial dyes as the tactic most likely to appear in incidents.
Method: multivariate anomaly detection across price stress, yield shocks, seasonality, historical fraud modes and adjacent-commodity analogues. Both pairings came out high likelihood and high impact — betel nuts to artificial sweeteners, with cyclamate as the analyte; orange juice to artificial dyes, with Sunset Yellow FCF (E110) as the analyte.
What was recommended in June, before the incidents?
- Targeted analytics. Add cyclamate to incoming betel nut test panels; add E110 to orange juice screening.
- Supplier controls. Request batch-level process records, verify formulation and colourant declarations, and scrutinise sudden yield or mass gains and Brix anomalies.
- Sampling strategy. Focus on higher-risk origins and suppliers, and on lots with abnormal specifications — unusual sweetness, unusual colour intensity.
Naming the analyte is the whole of the advice. A screen that looks for everything finds what it happens to cover; the same approach applies whenever a commodity has a known adulteration economics, as with the methods set out in our guide to detecting honey adulteration.
Knowing a commodity is at risk is the easy half.
The useful half is knowing which adulterant to put on the panel, and when. iComplai ranks tactic against commodity continuously, so the analyte is chosen before the incident curve turns rather than after.
See it on your own ingredientsWhat did the incident curves do?
They tracked the June predictions. Notifications involving cyclamate in betel nuts rose from a baseline average of 17 to 103, an increase of 605%. Notifications involving Sunset Yellow FCF in orange juice rose from 12 to 94, an increase of 783%. Both surges landed in the commodity and on the tactic that had been named six months earlier.
What generalises?
- Tactic-level forecasting multiplies yield. Predicting which adulterant to test for — sweetener rather than dye — finds more than a broad unguided panel with the same budget.
- Stress indicators lead. Harvest loss, disease pressure and extreme weather precede fraud upticks. Building those drivers into the model is what produces the lead time.
- Specificity speeds response. A named analyte, cyclamate or E110, lets inbound QA act immediately instead of scoping a method first.
The prediction anticipated the tactic and gave a concrete testing target months before the incident peak. The 605% and 783% rises are what confirmed it, and they are also the reason the same method was already running on formaldehyde in Chinese cabbage a year later.
Food fraud FAQ
Why predict the adulterant rather than just flag the commodity?
Because a commodity flag tells a QA team to worry and nothing more. An analyte tells them what to order. Naming cyclamate for betel nuts and E110 for orange juice turned a general fraud warning into a line on a test panel, which is the point at which a prediction starts costing less than it saves.
What made these two commodities predictable in June?
Price stress and supply shock. Betel nuts were under harvest loss and market tightness; orange juice was under high prices, disease pressure and extreme weather. Those drivers historically precede fraud upticks, so they carry lead time in a way incident data cannot.
How large were the increases, and over what period?
Between July and December 2025, notifications involving cyclamate in betel nuts rose from a baseline average of 17 to 103, a 605% increase. Notifications involving Sunset Yellow FCF in orange juice rose from 12 to 94, a 783% increase.
The commodity is the easy half. The analyte is the useful one.
See which tactics iComplai is ranking against your own commodities, in a 30-minute walkthrough that uses your ingredient list.
Talk to iComplaiWhere these figures come from
Risk-index readings, alert dates and the notification counts behind the 605% and 783% figures are iComplai platform records.