This paper (open access) proposes an approach to deal with surface scattering in the Near-Infrared (NIR) analysis of particulates.  The authors use coffee cultivarl testing as an example.

Surface scattering is a major confound in near-infrared (NIR) analysis of particulate foods. In roasted coffee powders, inhomogeneous scattering can obscure cultivar differences. Standard practice eliminates scattering (extended/multiplicative scatter correction) via reference-anchored polynomial projection.  The problem is that in heterogeneous matrices this is order-sensitive and can remove analyte-relevant variance.

The approach proposed in this paper is to encode scattering explicitly.  Per-spectrum polynomial slope, curvature and cubic baseline coefficients are determined and appended as descriptors and models are trained on the augmented matrix.

The authors reported that, using 300 diffuse-reflectance FT-NIR spectra (10 000–4000 cm⁻¹) from 25 lots covering 7 Arabica cultivars, this strategy improved test-set authentication.  Linear Discriminant Analysis (LDA) increased from 70.8% to 83.3%; Support Vector Machine (SVM)-linear from 63.9% to 84.7%, while Quadratic Discriminant Analysis (QDA) remained high (88.9%).

They conclude that the approach provides a quantitative scattering-aware method, treating it as information rather than aiming for its blind elimination. This method is an interpretable, easily implemented alternative and is applicable to other particulate and microstructured foods (e.g., cocoa, tea, spices)

The European Commission launched a new artificial intelligence (AI) platform on 10 March, TraceMap, to accelerate the detection of food fraud, contaminated food and foodborne disease outbreaks across the EU. TraceMap is accessible to national authorities in all Member States,.

TraceMap will use AI to:

• Improve food safety risk assessments by streamlining access and analysing critical data.
• Rapidly identify links between operators and consignments.  
• Monitor the entire agri-food supply chain, once a risk is identified, enabling faster recalls of unsafe or fraudulent products.

The intent is to enable national authorities to better target controls and carry out more thorough investigations, without requiring additional resources. It will use the extensive data in the existing EU agri-food systems to track trade patterns and production flows. The platform will improve screening accuracy, speed up the detection of suspicious operators and help investigators to detect food fraud and food borne outbreaks and remove non-compliant products from the market quickly. It will  enable better control of imported goods, in line with the strengthened measures set out in the Vision for Agriculture and Food.

TraceMap has been created by the Commission, using AI technology that processes, structures and interprets data from different food safety management platforms across the EU, including the Rapid Alert System for Food and Feed (RASFF) and Trade Control and Expert System (TRACES). A pilot version of TraceMap was recently used to support the identification and recall of infant milk formula made with contaminated ARA oil from China.

Photo by Mario Verduzco on Unsplash

In this study (open access) researchers developed and validated new new polymerase chain reaction (PCR) systems for detection of rapeseed in small samples of highly processed vegetable oils oils.

They designed primerss targeting the rapeseed acetyl-CoA carboxylase (ACCase; BnACCg8) gene, and optimised the PCR conditions after genomic DNA extraction from ground seeds and 700 µL aliquots of edible oils. DNA was isolated using two commercial kits, and PCR products were assessed by agarose gel electrophoresis.

They report that Uniplex PCRs demonstrated species specificity, producing 147-bp and 174-bp amplicons only in rapeseed DNA, with no amplification in soybean, sunflower, or maize. PCR bands from oils were weak or absent but implementing a double-PCR approach increased detection sensitivity in oils by approximately fivefold. Strong, expected-size amplicons were obtained from all oil extracts, confirming reliable detection of rapeseed in both cold-pressed and refined varieties, regardless of extraction method.

They conclude that this approach offers a sensitive, rapeseed-specific molecular tool for verifying the botanical origin of edible oils. It is suitable for routine authenticity testing and quality control of vegetable oils.

It is very difficult to verify, by testing honey, whether bees have been fed with C3-derived sugar syrups during the foraging season.  Sugar-feeding is not permitted unless over winter to keep the bees alive.

 In this study (open access), the researchers set out to show the potential for discrimination of sugar-fed hives using the non-exchangeable hydrogen isotope ratios on ethanol derived from honey, measured using mass spectrometry (ethanol isotope ratios are the same discriminator that underpin the proprietary SNIF-NMR databases that have been accepted for many years for fruit juice authenticity testing and have also been applied to honey)

 To generate reference samples, 36 genetically similar bee colonies, at a single geographical location and time point, were subject to different controlled sugar feeding regimes.  Four different sugar syrup types were used to represent distinct adulteration scenarios: fructose and glucose syrups derived from C4 plants, invert sugar derived from C3 plants (sugar beet), and sucrose syrup of unknown botanical origin.  Controls were in place to stop the colonies cross-feeding.

 The authors report that Ethanol δD_n values for adulterated samples differed significantly from controls, enabling clear discrimination.  This discrimination could form the basis of a potential classification database.

The application of high-throughput sequencing (HTS) of DNA barcodes can be hampered by technical challenges particularly in highly processed food.  Food pre-processing and differences in guanine/cytosine composition can lead to unequal amplification or complete loss of DNA barcode components.

To address this, the authors of this study (pre-publication, open access) used a multi-omic approach that coupled DNA barcode HTS analysis with proteomic analysis.  They applied it to the authentication of herbal beverages.

To resolve discrepancies between genomic and proteomic findings, the authors used traditional botanical morphology as an arbiter.

They applied their approach to a survey of herbal teas on the market in Russia.  They report two adulterations of Epilobium with Lythrum — a substitution potentially hazardous to consumers (Epilobium species are popular botanical drinks around the world, including Rosebay Willowherb, “Fireweed” and “Ivan Tea”).  They also found several minor substitutions, all confirmed by orthogonal methods.

They conclude that proteomic analysis provides enhanced confidence for verifying the presence or absence of plant components identified by HTS. However, its effective application is guided by prior sequencing to define specific targets for subsequent proteomic verification. A multimodal analytical approach is not only beneficial, but essential for the reliable and comprehensive characterization of components in complex plant mixtures.

Photo by Tamara Harhai on Unsplash

DNA-based verification that gelatin-containing foods and cosmetics do not contain pork products has always been a challenge due to DNA damage and destruction during gelatin production. 

In this study (open access) the authors report that – by careful optimisation of conditions – they could successfully apply a “traditional” PCR test to the problem.

They describe DNA extraction, post-isolation DNA analysis, annealing temperature and primer concentration optimization, specificity assay, amplification efficiency trial, sensitivity test, repeatability examination, and marketed sample analysis.

They report that the developed method demonstrated good specificity under optimized conditions. It achieved a good amplification efficiency of 101.2% with an R² of 0.994. The real-time PCR technique had a limit of detection of 1,316 pg in the sensitivity examination and a coefficient of variation of 0.81% in the repeatability testing.

They tested 10 retail samples (five facial mask cosmetics, food additive gelatin powder, two marshmallow products, and two gummy candy products), reporting that all of the samples displayed no amplification and were thus considered not to contain porcine DNA, consistent with the manufacturers’ labels.

The authors conclude that their real-time PCR method meets the validation criteria for qualitative analysis, including specificity, amplification efficiency, sensitivity, and repeatability.

Here is our latest monthly graphic from the EC Reports of Agri-Food Fraud Suspicions, showing a rolling 3-month trend. 

Our interpretation of the reports is subjective. In order to show consistent trends we have excluded cases which appear to be unauthorised sale but with no intent to mislead consumers (e.g. unapproved food additives, novel foods which are declared on pack), 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 highest proportion of fraud continues to relate 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.  Beyond this, there has been a steady increase in examples of foods labelled as "preservative free" being found to contain preservatives.  There are also consistent - and high - numbers of the simple fraud of food being under the declared pack weight.  A high proportion of these cases are due to excessive water glaze on frozen seafood.

A small - but increasing - proportion of the cases of non-meat/fish products being low in a premium component relate to "nutraceuticals"; foods marketed on the basis of an ingredient or additive with a real or implied health benefit, and the ingredient/additive being at a lower quantity than declared.  For those formulating such products, it is a reminder that there is no "under-tolerance" in the enforcement guidelines for declared amounts of such ingredients (including vitamins and minerals) and that the declared quantity must remain valid throughout the shelf life.

These Agri-Food suspicions are 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 multiple sources.

• JRC – These are solely media reports.  They exclude cases not in the public domain, and can be biased by shocking but highly localised incidents in local food supply within poorly regulated countries.  They now incorporate a search and trending tool to produce graphs and charts
• EU Agri-Food Suspicions – These are solely EU Official Reports, and only suspicions.  The root cause of each incident is unknown.  The data include pesticide residues above their MRLs. unapproved supplements and novel foods, and unapproved health claims.
• Food Industry Intelligence Network Fiin SME Hub – These are aggregated anonymised results from the testing programmes of large (mainly UK) food companies.  The testing programmes are targeted and risk-based, not randomised, and the fraud risks within such suppliers of large BRC-certified retailers and manufacturers may be different than the companies supplying small manufacturing businesses or hospitality firms.  

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 commercial providers, which gives very high level smoothed data based on official reports.

Wagyu beef's distinctive flavour and tenderness arise from its high levels of unsaturated intramuscular triglycerides. Although these compositional distinctions provide a unique Raman signature, extensive band overlap and background attenuation from packaging and frozen conditions hinder reliable in situ classification and constrain the interpretability of raw spectra.

This study (USD36 download fee) presents a label-free Raman spectroscopic and chemometric approach for authenticating Wagyu beef under realistic retail-like conditions. A supervised partial least squares discriminant analysis (PLS–DA) model was developed using spectra from unwrapped adipose tissue and evaluated with an independent validation set of frozen, plastic-wrapped samples from multiple breeds and suppliers.

The authors report that the model achieved 100% sample-level classification accuracy, To elucidate the molecular basis of discrimination and resolve spectral congestion, they used a two-stage decomposition combining singular value decomposition (SVD) and nonnegative matrix factorization (NMF), followed by nonnegative least squares (NNLS) fitting with pure triglyceride standards.

They found that analyses yielded chemically interpretable components, revealing enrichment of unsaturated triglycerides in Wagyu beef consistent with established compositional data. This could form the basis of a non-destructive test applicable directly through plastic packaging.

Photo by moreau tokyo on Unsplash

Undeclared lard in confectionary products is a significant concern for consumers in many parts of the world who avoid pork on religious grounds.

This study (GBP30 download fee) used gas chromatography with flame ionization detection (GC–FID) to measure fatty acids, then principal component analysis (PCA) to detect porcine fatty acid biomarkers in imported chocolates and biscuits.

The authors report that total fat content ranged from 11.5 to 32.5%, with palm kernel-based chocolates enriched in lauric (42–52%) and myristic acids (18–20%), while other chocolates were dominated by palmitic, stearic, and oleic acids. Biscuits contained high proportions of palmitic and oleic acids (> 75%).

PCA of the complete fatty acid dataset separated lard-adulterated samples.. Targeted PCA using porcine biomarkers palmitic-to-oleic acid ratio and eicosadienoic acid confirmed this clustering.

Calibration using simulated lard–palm oil mixtures (0–15% w/w; five replicates per level) enabled quantitative estimation of lard .

Photo by Pawel Czerwinski on Unsplash

Raman, and related techniques, have the potential to provide field-based rapid and non-destructive testing for dairy products and powders.

This review (open access) consolidates advances reported from 2015 to early 2025, covering conventional Raman, surface-enhanced Raman spectroscopy (SERS), Fourier-transform Raman, hyperspectral Raman imaging, confocal/mapping approaches, and portable systems.

The authors critically evaluate preprocessing and chemometrics as well as machine-learning and deep-learning pipelines for classification and quantification.

They compare species-specific applications including cow, buffalo, goat, camel, donkey, human breast milk (macronutrients, sex-linked profiles, microplastics, antibiotics), and milk powder workflows with respect to matrix effects, fluorescence interference, and validation practices.

They summarise that  Raman enables chemically specific fingerprints of proteins, lipids, and carbohydrates, whereas common adulterants present diagnostic bands. SERS substrates routinely extend sensitivity to ppm–ppb levels and suppress fluorescence, supporting rapid detection of melamine, urea, ammonium sulfate, thiocyanates, benzoate, and selected antibiotics. Hyperspectral imaging provides spatially resolved maps, differentiating multi-adulterant mixtures and thermo-structural behavior in powders.

Chemometric models achieve high accuracy for classification and concentration prediction, whereas deep-learning architectures improve robustness under nonlinear matrix variation and instrument drift.

They conclude that challenges persist in substrate reproducibility, calibration transfer, fluorescence in lipid-rich systems, and detection of emerging adulterants and trace preservatives under field conditions. Future progress will hinge on multi-excitation instruments with adaptive laser power control, universal SERS substrates integrating plasmonic metals, dielectric shells, and molecular recognition, and standard operating procedure grade preprocessing. They highlight that industrial reliability requires calibration-transfer strategies, rigorous validation, and explainable artificial intelligence to link decisions to chemically meaningful features, supporting regulatory acceptance and auditability.

Portable Raman and SERS systems can aid nutritional profiling and contaminant surveillance in breast milk, whereas Fourier-transform Raman and hyperspectral imaging mitigate fluorescence and map heterogeneity in powders.

Verifying the origin of garlic has risen up the risk rankings in recent years.  Approximately 70% of the world’s garlic originates from China.  Volatility in trade tariffs (and the anticipation of tariffs) and anti-dumping measures mean that there could be financial incentive to trans-ship Chinese garlic through a third country and mis-state the country of origin, particularly if importing into the US.

In this proof of concept study (USD32 download fee) the authors show that microbiota profiling provides an alternative to conventional chemometric approaches for garlic origin authentication. They characterized the surface bacterial communities of 153 garlic samples collected between 2021 and 2024 from China (n = 60), the United States (n = 50), and multiple other countries (n = 43) using 16S rRNA gene amplicon sequencing.

They report that comparative analyses revealed significant differences in alpha and beta diversity across countries, with U.S. samples exhibiting the highest microbial richness and Chinese samples the lowest. Dimensionality reduction methods showed clear clustering by country of origin, supporting the presence of distinct microbial signatures. Machine-learning classifiers trained on 16S profiles achieved >0.87 accuracy across Random Forest, k-nearest neighbours, logistic regression, and support vector machine models using only five genus-level microbial features.

Multi-year sampling confirmed that these microbial signals remained stable across harvest seasons. Differential abundance analyses further identified ecologically relevant taxa driving country-level separation.

Photo by team voyas on Unsplash

This technical report from the FAO (free to download) provides a detailed overview of food fraud in the aquatic sector, outlining its types, causes and impacts. It demonstrates that species substitution and mislabelling are the most common forms of fraud, with studies indicating that up to 20 percent of fishery and aquaculture products globally are mislabelled. Fraud is especially prevalent in restaurants and catering services, where visual identification is challenging, and in processed products, where the species identity can be masked.

A series of international case studies illustrates the extent and consequences of food fraud in the aquatic sector and provides an overview of the most common cases and the available tools to fight food fraud in the sector.

The report reviews international regulatory frameworks and standards designed to mitigate fraud risk, including Codex Alimentarius, FAO guidelines, and GFSI‑benchmarked schemes (such as BRCGS, FSSC 22000, International Featured Standards, and Safe Quality Food), as well as national laws in Australia, Canada, the United States of America and the European Union.

It advocates for harmonized labelling requirements, the mandatory inclusion of scientific names, and better traceability systems. Raising consumer awareness and

increasing industry transparency are also highlighted as critical steps to reduce fraud and support sustainable practices in the aquatic sector.

The report also includes summaries of the most common testing methods used to identify different types of fraud.

This study (open access) used machine learning classification models to identify monosaccharide markers for coffee adulteration.  These markers (proposed thresholds for glucose, xylose and mannitol) are suitable for authenticity monitoring vs Brazilian official regulatory standards (SDA Ordinance 570) using High-Performance Anion Exchange Chromatography with Pulsed Amperometric Detection and can flag adulteration with corn, wheat, and barley adulteration from 3%.

The training and validation sets were prepared from verified samples supplied by the Brazilian Ministry of Agriculture and roasted, ground and adulterated in-house.  Coffees (157 raw samples) comprised of arabica and canephora species from eight different states.  Adulterants were acai, husk, barley, wood fragments, corn and wheat ranging from 1 – 20%.

Photo by Nathan Dumlao on Unsplash

Near-Infrared (NIR) sensors are routinely used for in-process monitoring in the cheese industry, from raw milk analysis to final product grading. For example, in curd processing, real-time NIR monitoring of moisture and fat content enables dynamic adjustments to cutting and cooking parameters, reducing batch inconsistencies.  During ripening, hyperspectral NIR imaging tracks proteolysis and lipid oxidation, providing insights into flavour development and shelf-life prediction.

There have been many proof-of-concept studies to extend the technique from quality monitoring and in-process adjustments to real-time checks for authenticity or chemical contaminants.  None have yet made it into routine use.  This review (open access) discusses the current gaps, the latest developments, and argues that – with the pace of AI development – these gaps could soon be closed, particularly in the PDI/PGO cheese supply chains.  Success would require coordinated efforts among research laboratories, regulatory authorities and producers to establish harmonised protocols, shared spectral repositories and validation frameworks.

This study ($25 download fee) compiled 254 incidents of food adulteration reported across from 20 countries in the Middle East and North Africa (MENA) region between 2019 and 2024, gathered from primary sources published in Arabic (85 %), English (10 %), and French (5 %). It also analysed 1261 notifications from the RASFF concerning food products originating from MENA countries during the same period.

The authors report that Lebanon and Turkey contributed the highest number of reported incidents with mislabelling (particularly expiry-date falsification) being the most common fraud.

The web-based surveillance identified 254 incidents, with Lebanon contributing to the highest number (15 %) followed by Egypt, Jorda and Iraq, while 78.9 % of all signals were classified generically as “food product’ and the most common issues involved expiration-date manipulation (62.9 %).

In the RASFF system, 1261 notifications linked to MENA-origin products were recorded, dominated by Turkey with 564 notifications (44.7 %) followed by Egypt (18 %) with alerts increasing between 2019 and 2024 and mainly triggered by contaminants (45.7 %) or unauthorized substances (16.9 %)

[Image – EverythingBen, available under Creative Commons Universal Public Domain Dedication]

Milk adulteration remains an endemic problem in many regions of the world.  Police action, consumer illness and even fatalities are reported regularly from countries such as India and Pakistan.  There is a need for simple, low-tech, cheap, tests that can be used by either business customers or by the public.

This paper (purchase required) describes a low-cost hybrid paper/plastic strip test for the simultaneous detection of seven potential adulterants in cow milk: urea, hydrogen peroxide, starch, formaldehyde, antioxidants, sodium hypochlorite, and neutralisers/detergent.  It uses pH-based colorimetric sensing without sample pretreatment. The device was fabricated using a craft cutter and combined paper and plastic substrates, allowing multiplexed detection on a single strip.

The authors report that the test strips remained stable for up to 30 days under refrigeration (2–5 °C). In a case study of 50 milk samples, the device accurately identified adulterants with minimal interference, and results showed no significant deviation from reference methods.

Fabrication costs are around $0.25 per unit.  The authors conclude that the proposed platform provides a reliable, affordable, and scalable solution for routine milk quality monitoring, representing a promising tool for enhancing quality control in the dairy industry.

[image from the publication]

The European Commission’s Joint Research Centre (JRC) have recently produced “authentic” and “inauthentic” honey reference materials.  These materials will shortly be available within laboratory proficiency testing (PT) schemes.  Analysis of reference materials (where a laboratory’s result can be compared to a traceable known value) and participation in PT schemes (where laboratories compare their own results to those of their peers testing the same sample) are critical aspects of the Quality Assurance system of any testing laboratory, and are mandatory where testing is accredited to ISO 17025.

Participants in LGC AXIO’s Food Chemistry proficiency test for honey parameters will have the option to take part in an additional exercise in the June 2026 round. This add‑on focuses on the analysis of honey authenticity.  Alongside the standard honey sample, laboratories may choose to analyse two extra honey samples supplied by the JRC.  These materials are intended to support the interpretative assessment of sugars commonly used as authenticity markers.

The two supplementary honey test materials are intended for:

• Sucrose analysis
• Fructose analysis
• Glucose analysis

Participants will also be asked to provide an authenticity assessment for each sample, including whether it meets the laboratory’s own criteria and the basis for that judgement.

 Laboratories that would like to receive the additional honey samples should contact axiopt@LGCGroup.com 

 LGC AXIO are a partner of FAN and we continue to be grateful for their support alongside all our partners.

Photo by With Mahdy on Unsplash

It is difficult to distinguish fresh from defrosted lamb using a single analytical test.  In this study (open access) the authors propose a screening approach using Near Infra Red spectroscopy (set up online in a production environment) followed up by a panel of classical laboratory tests if required for further investigation including pH, colour parameters (L*, a*, b*), lipid oxidation (TBARS), cooking loss, and Warner-Bratzler shear force.

They used machine learning, feature selection and multivariate statistics to build classification models for each test.  The models were trained on samples from twenty crossbred lamb carcasses from various commercial butcher shops. The animals were intentionally sourced from different regions of Bangladesh to ensure genetic and environmental diversity among the samples. A total of 400 meat samples were collected from these 20 carcasses, with five anatomical cuts, loin, round, rack, leg, and breast, taken from each carcass. All samples were immediately placed in sterile, ice-filled containers and transported to the laboratory then evenly divided into two groups: 200 for fresh condition analysis and 200 for frozen condition analysis. All samples were first stored at 4 °C for 24 h to allow proper post-mortem muscle-to-meat conversion. After the chilling period, the fresh group was analyzed immediately, while the remaining 200 samples were stored at −20 °C for 30 days to represent the frozen condition.

The researchers report that classification models could be built using the “classical” laboratory tests alone but they introduced the risk of overfitting.  When an NIR classification model was added to the workflow as an initial screen this provided a more robust analytical approach.

[image from the publication]

This paper (purchase required) reports the comprehensive analysis of 437 standardized wine samples produced from six distinct grape varieties over a span of two decades (2002–2023).

The authors report that the grape variety was pivotal in shaping the composition of phenolic acids, flavan-3-ols, and anthocyanins. In total, 27 parameters were examined. The grape varieties Blauer Wildbacher and Blaufränkisch exhibited the most significant differences when compared to other red wine varieties. Furthermore, the phenolic profile showed the relationship between the varieties. The phenol content of the variety Zweigelt wines exhibited a stronger correlation with the vintages. The influence of the location on the phenolic profile could not be proven in most cases, except for the variety Blaufränkisch.

A comprehensive analysis encompassing all varieties, locations, and phenol analytes was conducted.  This revealed an increase in phenol concentrations over the period of vintages studied.

Photo by Kelsey Knight on Unsplash

FAN has a searchable index of where to find databases (either analytical signals or compositional parameters) of authentic food.  These are used as reference benchmarks for analytical authenticity tests.

We are in the process of updating this list.  As well as reference data sets for untargeted testing, which are typically held in-house by laboratories, we now include public datasets of benchmarked food composition; genetic data, lipid profiles, sugar profiles, aroma profiles, metals and minerals, composition of branded foods and many more.  If you know of a dataset that should be listed then we would love to hear from you.  Please contact secretary@foodauthenticity.global