The NFCU's industry updates highlight the key risks and issues that may be impacting the food industry, share best practice to strengthen the industry’s response to food crime and inform on the ongoing work of the NFCU.
In this edition:
Read the November Update here.
You can contact the NFCU Prevention team to feedback, raise a concern or possibly contribute to a future update.
This paper (open access) reports the development of a classification model for the geographic origin on black tea based upon measuring a panel of 15 trace elements by X-ray fluorescence (XRF). XRF is a non-destructive technique. The only sample preparation required is grinding the tea leaves into a fine powder.
The model could discriminate between 10 major tea-producing regions. It was built using reference samples obtained, via tea industry contacts, directly from plantations or primary processing facilities. 791 black tea samples were collected in total: Assam (272 samples), Burundi (40 samples), Darjeeling (145 samples), Ethiopia (40 samples), Keemun (115 samples), Kenya region 1 (41 samples), Kenya region 2 (40 samples), Malawi (40 samples), Rwanda (10 samples), and Sri Lanka (48 samples).
Two unsupervised analysis techniques were used to visualize high-dimensional data, and six supervised models were employed to discriminate the ten GI regions.
The authors report that machine learning models, including random forest, support vector machine, k-nearest neighbours, linear discriminate analysis, and the deep learning multilayer perceptron (MLP) model, demonstrated superior predictive capabilities compared to the traditional partial least squares discriminant analysis model. The MLP model achieved the highest performance, with a 97.7 % overall F1 score in predicting the geographical origins of 532 authentic samples across ten GI regions.
The authors also Identified Rb, Sr, Mn, Si, and Cl as geographical markers for African region discrimination.
The conclude that their work could form the basis and foundation for an international database of tea Geographic Origin, enabling cheap and quick authenticity verification testing.
Photo by Oleg Guijinsky on Unsplash
Artificial Intelligence (AI) is increasingly applied in food safety management, offering new capabilities in data analysis, predictive modelling, and risk-based decision-making.
A review of the literature identifies three primary areas of application: scientific advice, inspection and border control, and operational activities of food safety competent authorities.
Five country examples with the real-world use cases illustrate diverse uses of AI tools, including pathogen detection, import sampling prioritization, and language models for regulatory data processing.
Regulatory frameworks, as well as voluntary governance, addressing AI in the public sector are emerging worldwide. National and international initiatives often highlight the importance of data governance, transparency, ethical considerations, and human oversight. Challenges such as biased data, explainability, and data governance gaps appear across different contexts, along with potential risks from deploying AI systems prematurely. Access to high-quality, interoperable data and collaboration among stakeholders can support effective integration of AI technologies.
AI readiness often depends on understanding specific problems to be addressed, current capacities, and the quality of available data. Human oversight and continuous evaluation contribute to maintaining trust in AI systems.
Collaborative efforts involving academia, the private sector, and international organizations help build shared knowledge and resources for AI development in food safety. Overall, AI presents opportunities to enhance resilience, efficiency, and responsiveness in food safety systems. Careful consideration of governance, data management, and multi-stakeholder cooperation can shape AI’s contribution to achieving sustainable and equitable outcomes in agrifood systems.
Read full report here: https://openknowledge.fao.org/handle/20.500.14283/cd7242en
Injera is a dietary staple in Ethiopia, eaten with most meals. It is a flatbread made with teff flour. Injera is vulnerable to adulteration with cheaper gesso or cassava flours.
This paper (purchase required) reports a simple, affordable, portable, and easy-to-use method based on a paper analytical device to indicate adulteration qualitatively.
The authors report that the developed test card generated a red-orange colour on lane B (ferric detection), red on lane D (ferrous detection), Prussian blue on lane F (ferric detection), and Turnbull’s blue colour on lane H (ferrous detection) for pure teff injera. The colour barcodes generated by pure teff injera differ from those produced by teff injera that contain gesso or cassava.
In a survey of local market produce, the test card colour result was less intense or inactive in most cases. It indicates that inexpensive cereals might be used in place of authentic teff flour or flours have been blended before baking.
The authors cross-validated their method by analysing the elemental composition of samples using microwave plasma atomic emission spectrometers.
Photo by Syed F Hashemi on Unsplash
The vacancy is within the Unit JRC.4 “Food Integrity”, whose mission is to produce and validate the knowledge for ensuring authenticity, quality and sustainability of foods and to contribute to the fight against adulterated and illicit consumer products.
|
2025-JRC.F.4-GEE-FGIV-002375 |
|
Dec 05, 2025 23:59 Brussels time |
|
European Commission, Joint Research Centre, Geel, Belgium |
|
Auxiliary Contract Staff |
|
FGIV |
For complete information, please download the Vacancy Notice.
From Boardroom to Berlin: EFF-CoP in Action
For more than half a year, the EFF-Editorial Board has been steadily shaping the voice of a growing community. Through frequent meetings, the Editorial Board reviews ideas, discusses themes, and schedules each upcoming EFF-publication in the New Food Magazine.
Their collaborative rhythm has already produced three published articles, all available to registered users on the EFF-Hub, forming a reliable stream of insights for everyone across the food-fraud field. Read HERE the latest EFF-article.
That same spirit of shared expertise came to life at the International Food Fraud Conference 2025 in Berlin. EFF-CoP partners from across Europe gathered to exchange knowledge and spark new conversations. On the first day (5/11/2025), EFF-CoP led a workshop exploring the “state of the art and main challenges,” inviting participants to identify gaps and build a shared vision for safer, more transparent food systems.
The second day (6/11/2026) featured a joint session with the Watson Project, examining how climate change, geopolitical shifts, and economic pressures may reshape food fraud in the decade ahead. The day concluded with an uplifting presentation by EFF-CoP’s coordinator, Prof. Saskia van Routh, titled “The Power of We: The European Food Fraud Community of Practice Story,” in which she also introduced the Food Fraud Festival in Dublin (27–28 May 2026)—a future gathering where collaboration, innovation, and community will continue to grow, offering many insights into food fraud.
For more details, photos and videos, click HERE.
Finally, don’t forget to register on the EFF-HUB - the central meeting point for our food fraud community. There, members can chat, exchange knowledge, and access EFF-CoP’s materials, webinars, and workshop updates. Together, we’re not just learning about food fraud - we’re building the future of prevention, one workshop at a time.
This update has also been added to the FAN EFF-CoP page.
Food fraud prevention and detection priorities can be different in different countries. In Iran, as in many countries, pork meat is an unlikely adulterant in beef or chicken sausages as there is virtually no pork production; it is legally and culturally proscribed. Donkey and horse, however, is not food grade but is cheap and readily available as an adulterant. Laboratories with PCR are scarce and the need is for rapid, portable verification tests.
The researchers in this study (open access) sought to address this need by developing a classification model using non-destructive FTIR. They deliberately omitted any extraction or defatting step so that the test could be applied directly to a 3mm slice of the intact sample. They trained the model using sausages prepared in-house that mimicked – as far as possible – the typical recipes used in Iran (40 – 60% meat content, along with soy flour, egg, herbs and spices). They prepared 20-each of beef, chicken, horse and donkey sausages using meat sourced directly from veterinary schools. For the training set, triplicate sub-samples were measured from each sausage and then the triplicates averaged. Some pre-processing was applied to the data before dimension reductions using supervised machine learning. 30% of the samples were reserved as a validation set and kept independent of the training set. Additionally, within the training set, a 5-fold cross-validation procedure was used to iterate an internal check against over-fitting.
The researchers were able to separate the four species into distinct clusters using Principle Component Analysis. They also postulate a chemical rationale as to why these identified signals should differ between species. They conclude that their approach could form the basis of a rapid non-destructive test with practical application.
Image – from paper
Spain has a legal limit of 3% for undeclared vegetable proteins in meat patties. The aim of this open-access study was to evaluate the feasibility of point-based near infrared spectroscopy (NIRS) and hyperspectral imaging (HSI) to verify compliance.
The model was trained on patties prepared in-house. They were all prepared from the same cut of beef, so the robustness of the model has not been verified. A total of 240 patties were fabricated, of which 60 contained pea (PP), 60 contained soybean (SP), and 60 chickpea protein (CP) at levels from 1 up to 6 % (w/w). 60 pure beef patties were included.
The authors report that they could clearly discriminate the type of protein added, using either partial least squares-discriminant analysis (PLS-DA) or linear discriminant analysis (LDA), with >90 % of the samples in the test set correctly classified. Based on protein inclusion, LDA discriminated 100 % of the PP, SP and CP samples with both NIR and HSI. PLS-DA classified 100 % of the PP and CP burgers using the NIR instrument. To manage double classification tasks, a hierarchical model classifier (HMC) was proposed for both NIR and HSI spectra, achieving classification rates of at least 83% by combining LDA and PLS-DA models at the nodes.
The authors conclude that NIR spectroscopy is suitable for detecting low levels (1 %) of vegetable protein flours added to beef burgers.
This article (open access) discusses some of the current challenges in testing animal feed for compliance with European legislation using microscopy (one of the official methods mandated by the legislation).
Although the article is pre-publication and not peer-reviewed, it is generously illustrated with colour photographs of microscope slides - such as that shown here - which could be a valuable training aide to Official Control analysts who are relatively new to microscopy. The author discusses key areas where expert interpretation is needed, describing the examples of bovine spray-dried plasma protein (legal, within restrictions, in the US but banned in the EU in ruminant feed), differentiating milk powder from blood powder, and differentiating hydrolysed proteins from vegetable vs animal sources.
The recent proliferation of tariffs on internationally-traded food has led to an increased focus on origin claims. False origin claims can enable tariff avoidance. On-pack claims also resonate strongly with consumers in some countries as consumers push back against the politics of tariffs and they increasingly champion home-produced food. One of the clearest examples of this trend is in Canada.
There has been a renewed enforcement focus on “home-produced” claims in Canada. A blog by legal firm Blake Cassels & Graydon gives a good summary of the rules and links to the primary legislation
Classification models for food authenticity tests can - in principle - be based on any analytical technique that collects multi-variate data. In the case of spectrometric data (such as NIR or multi-spectral imaging) the equipment can be relatively cheap. For collecting chemical data, researchers often use high-end equipment such as advanced LC-MS or GC-MS
This proof-of-concept study (purchase required) is a rare example of building a classification model using a cheaper test (HPLC with fluorescence detection) to measure a chemical parameter. The authors prepared cold-pressed walnut and pumpkin seed oils adulterated with 0 – 50% of sunflower oil. They developed a classification model based on the concentrations of the four tocopherols (α-, β-, γ-, and δ-). They report that the model was capable of discriminating sunflower oil adulteration down to 2-3%.
This short communication describes a simple chemical-based colour test for detecting fake saffron. No details are provided in the abstract about the basis of the test, or whether it discriminates botanically-related adulterants such as safflower. We have not purchased the full article in order to review it.
The abstract describes it as a panel of 4 simple chemical tests which take around 30 minutes to perform and can be read by eye from a colour card. The attached image is from a correction (open access) to the original publication (which remains behind a paywall).
N-glycans are a class of biological compound that are chemically bound to proteins. They are generally stable to food processing and heating.
In this paper (open access) the authors investigated how N-glycans varied amongst three different fish species; red snapper, barramundi (Asian Sea Bass) and the potentially cheaper adulterant, tilapia. They measured N-glycan profiles using liquid chromatography with ion mobility and mass spectrometry (LC-IM-MS). They identified four N-glycan structures containing different degrees of O-acetylated sialic acids (O-Ac-Sias) as species-specific markers and found clear clustering based on their percentage abundance. This enabled a multi-class species classification model. They found that this clustering and classification model remained valid even after the fish had been cooked and processed.
They conclude that this approach could complement DNA testing when looking at heavily processed or manufactured food.
One approach to verifying that undeclared GMOs in raw materials are compliant with legal thresholds is to test for GMOs in the manufactured product and then extrapolate the GMO content in the ingredient using the pro-rata recipe proportions. This is an attractive approach for enforcement testing which does not always have access to the raw materials.
The authors of this paper (purchase required) investigated the inherent bias in this approach. They prepared in-house model processed foods (heat-treated soybeans) containing GM events and then tested them using a GMO quantification method incorporating taxon-specific real-time PCR with longer amplicons They observed that the extrapolated GMO content increased with the length of the amplicon used in the taxon-specific PCR assay. When a longer amplicon was deliberately employed, the GMO content calculated for the processed food always exceeded that expected by pro-rata calculation from the raw material.
They conclude that this finding can be used to advantage. The use of longer amplicons in taxon-specific PCR can lead to an overestimation of GMO content at the raw material stage based on the measurements from processed foods. If the overestimated value remains below the labelling threshold, the appropriateness of GMO labelling can still be confirmed. The proposed method offers a simplified and practical screening approach for use in routine inspections.
The EU Joint Research Centre (JRC) have now published their monthly collation of fraud media reports for July 2025 and September 2025 (these collations are published retrospectively, and August’s report was published in advance of July’s). The full index of reports can be found here
These new reports have also been added to the JRC database that underpins a searchable front-end for media reports of food fraud incidents. It allows filtering by commodity, country, fraud type and other key criteria.
The JRC collation is just one of the incident databases available. Different databases collect different information, in different ways, and therefore show a different angle on the true picture. All of these sources are signposted on FAN. Best practice is to use a combination of all sources, but the final critical question is “how vulnerable is my own supplier”.
Many testing laboratories also supply their own customers with incident collations, and there are many commercial software systems that scrape reports from the internet. All collect and treat the data slightly differently. FAN produce a free annual aggregate of "most adulterated foods" from three of the largest commercial providers, which gives very high level smoothed data.
This project, funded by the UK Food Standards Agency and conducted by Fera Science Limited, aimed to identify and review current and emerging methods to detect adulteration in edible oils, focusing on issues relevant to UK consumers and the economy. The study involved a comprehensive literature review, stakeholder engagement, and consultation of proficiency testing and Fera Science’s HorizonScan™ data to assess future risks.
The review covered rapid screening methods, mainly spectroscopic, and confirmatory techniques such as fatty acid and triacylglycerol profiling.
The authors report that many approaches are still under development and lack thorough validation. A key challenge is the increasing sophistication of fraud within the supply chain, with businesses often relying on proprietary protocols, which hampers standardisation.
The report recommends addressing the lack of standardisation and regulation in edible oil testing, investment in widespread testing and point-of-use methods, and developing confirmatory techniques. Spectroscopy methods like Fourier Transform Infrared and Raman show promise for rapid, low-cost testing, while triacylglycerol analysis could serve as a confirmatory method for laboratories. Authentic certified reference materials are also essential to support quality control and encourage proficiency testing uptake.
A link and signpost to this report has been added to FAN’s Research Reports index.
This paper (open access) reports the development of a hand-held device that can detect methanol addition in alcoholic spirits by scanning directly through the unopened glass bottle. Such a device has obvious benefit to enforcement inspectors at ports and retail outlets. The paper also describes the operating principle of the device, including all the modifications made by the authors and why they were needed, in clear language understandable to non-specialists.
For an overview of Raman spectroscopy see FAN’s method explainers.
The authors of this paper describe the three main challenges to overcome in order to make a practical Raman Spectroscopy scanner which can read through glass bottles; 1) the spectroscopic signal from the container masking the sample signal; 2) the intrinsic fluorescence signal of the sample that can overwhelm the weaker Raman peaks; and 3) the opacity and colour of the glass attenuating the signal both entering and exiting the container.
They use of a combination of approaches to circumvent these challenges. They use an axicon lens to generate a conical excitation beam, which effectively circumvents the bottle signal. They also use a relatively long-wavelength excitation combined with wavelength modulation (Wavelength Modulated Raman Sprectroscopy, WMRS) to minimise and then offset any natural fluorescence from components in the drink.
To quantify, they compared the signals attributable to methanol with those from ethanol as an internal standard. They used the nominal %ABV of ethanol for this calculation, on the assumption that adulterated spirits would have a lower than declared ethanol content and therefore they would over-estimate the methanol content (i.e. erring on the side of caution, for a screening test).
They report the successful detection of methanol adulteration at well below the 2% level that causes acute serious health concerns. The method has been validated on one real spirit sample but has yet to be tested for robustness over a range of samples.
Image from the publication
Here is our regular monthly graphic from the EC Reports of Agri-Food Fraud Suspicions, showing a rolling 3-month trend. These EU reports are a useful tool for estimating fraud incidents, signposted on FAN’s Reports page. They can be found here.
Our graphical analysis contains some subjectivity in the interpretation of the report data. In order to show consistent trends we have excluded cases which appear to be unauthorised sale but no intent to mislead consumers of the content/ingredients of a food pack (e.g. unapproved food additives, novel foods), we have excluded unauthorised health claims on supplements, and we have excluded residues and contaminants above legal limits. We have grouped the remaining incidents into crude categories. Our analysis is intended only to give a high-level overview.
The absolute count of incidents are creeping up a little but are generally fairly steady. The split of incidents between different categorisations is also fairly consistent over time, with a significant number relating to falsified or unlicenced trade in high risk food (illegal operators, missing or falsified health certificates, attempts at illegal import) and relating to falsified or missing traceability documentation.
Another consistent theme is underweight premium ingredient content in processed food, generally (but not always) the meat or seafood content. Often this is associated with excess glaze or water in frozen food.
Although we do not count unapproved (but declared on-pack) additives in these graphs, it is noteworthy that there has been a consistent rise in recent months in the number of regulatory siezures of food (often confectionary) imported into the EU that contains titanium dioxide (an additive permitted in many regions of the world, but now banned in the EU). This increased enforcement activity may account for some of the general insight reports, based on analysis ot EU Agri-Food Suspicions, that "food fraud incidents are increasing".
The EC Monthly reports are only one source of information. A comparison of the many different information sources now available, and the complementary insight that can be gained from using a variety of information sources, is given in an earlier blog this year.
Species identification in canned tuna is much more challenging than for processed fish in most cooked foods. This is because the DNA is substantively degraded during the canning process.
In this paper (purchase required) the authors present a protocol to increase concentration and purity of DNA extracted from canned samples. The experiment mainly consists of: (1) drying the canned tissue in paper filter, (2) washing it with a PBS solution, (3) store in ethanol 96 % at −20°C, and (4) perform DNA extraction.
They report that the pre-treated samples showed an increase of both DNA concentration and purity indicating that some of the inhibiting molecules were successfully removed. These differences between the two treatments were statistically significant (p < 0.01). At the amplification level, the pre-treatment allowed the recovery of complete fragments of the barcode region COX1 with approximately 650 base pairs.
The authors recommend their approach should be used in combination with other methodologie such as mini-barcoding.
Photo by Grooveland Designs on Unsplash
This research (purchase required) set out to design a rapid point-of-use test to detect cassava starch as an adulterant in higher-value starches. The test method used Proofreading enzyme-mediated probe cleavage (Proofman) coupled with ladder-shape melting temperature isothermal amplification (LMTIA). The optimal detection temperature of this Proofman-LMTIA method was 62℃ and the reaction could be finished within 20 minutes with a detection sensitivity of 100 pg/μL of genomic cassava DNA. Nine different species were collected and verified for the specificity of cassava ITS primers and probes. The detection limit of cassava DNA derived from artificially premixed starch powders was 1 % (w/w).
As a proof-of-concept, the researchers used their Proofman-LMTIA assay to test 33 commercial products from the food and medicine sectors. They report that – based on their assay - 16 samples contained undeclared cassava components.
Photo by Daniel Dan on Unsplash
