News

The Government Chemist Annual Review for 2023 was presented to UK Parliament by the Parliamentary Under-Secretary of State for AI and Digital Government by Command of His Majesty. It was ordered by the House of Commons to be printed on 5 June 2025.

The Government Chemist Annual Review provides a summary of the work undertaken by the Government Chemist team, including highlights from the resolution of referee cases, advisory work and capability building activities. It includes work related to the in relation to food authenticity and safety, and includes an update on FAN. The review also details the impact of the work obtained though active engagement with a wide range of stakeholders.

Read full review.

The authors of this study (open access) used the results and datasets from 18 published projects and biobanks to build a database of bacterial metataxonomic data from fermented table olives.  The collated database contained database 442 samples of 16S rRNA bacterial profiles

They then compared three tree-based Machine Learning algorithms—Classification and Regression Tree, Random Forest (RF), and Extreme Gradient Boosting— to classify the origin or production process of the olives. They report that Machine Learning techniques can effectively classify bacterial profiles based on olive processing type, cultivar, country of origin, and isolation matrix. The Random Forest model achieved the highest accuracy, reaching 97% in the best cases, with a kappa coefficient above 0.8 for most categories.

They conclude that approach holds potential applications in the table olive sector and in other food products, where the industrial application of ML techniques to metataxonomic data could enhance traceability, authenticity, and quality control.

Photo by Melina Kiefer on Unsplash

The EU has updated its list (Delegated Regulation 2025/1184) of countries in which it considers controls against money laundering or terrorist financing are poor.  Any European business dealing with businesses in these countries is expected to enhance their financial due diligence checks and internal governance for contract review and sign-off.

Due diligence checking of new (and existing) suppliers is an essential fraud mitigation tool for any business, including food businesses.  The overall level of regulatory control in a given country, along with generic cultural attitudes to bribery and corruption, will inform this risk scoring.  A data source used by many businesses is the Corruption Perceptions Index league table published annually by Transparency International.

We have launched a new resource page (see 3rd tab) on our Food Crime webpages in order to collate estimates of the economic cost of food fraud.  If you are aware of other studies we can add to the three already cited then please let us know and we will add them to the page.

Estimates are inherently uncertain but the numbers are staggering.  Typically an annual $30 - $50 billion USD globally.  And that is just economic cost.  It is important to remember that food fraud often has a cost to human health or even life.

 Photo by Hossein Fatemi on Unsplash

Legal identity standards are an important benchmark for food authenticity.  They are set at a national (or EU) level, and can define everything from the minimum cocoa solids for chocolate to be branded as “chocolate” to the definition of a “meat pie”.  The difference between identity standards in different jurisdictions is a regular reason why internationally-traded foods are rejected as inauthentic. 

There are opposing trends in different parts of the world in terms of the scope and breadth of legal identity standards.  The EU has a wide range of Protected Geographic Indications, Protected Designations of Origin, and minimum specifications for many common foods.  The “Breakfast Directives”, covering jams, honey, fruit juices and milk, were tightened in 2024.  In the US, as part of the current national drive for deregulation, the FDA have just revoked 52 identity standards.  These mainly relate to foods that are seen as obsolete, or which have a very small commercial market within the US.  Meanwhile in India, where the scope of identity standards has been seen as narrower, the FSSAI have just tightened the norms for oils, sausages and colours.

With this continually evolving landscape, it is imperative that an exporting food company understands the identity standards of the territory where they are intending to sell.

See blog by legal firm Hogan Lovells on US deregulation of identity standards

See press report on India tightening of identity standards

Photo by Elena Leya on Unsplash

This work, originally presented at an Institute of Electrical and Electronic Engineers conference and now published in an IEEE journal (purchase required), provides an example of how a small team of researchers can develop a bespoke digital traceability system for the Agri-Food industry.  This provides an alternative approach to buying one of the distributive ledger systems available from large commercial software vendors.

The researchers developed a decentralized system for the agrifood supply chain that allows product traceability and quality assurance. System decentralization and privacy preservation were enabled through the combination of Self-Sovereign Identity (SSI), Decentralized Identifiers (DIDs), and Verifiable Credentials (VCs). DIDs provide stakeholders with complete control, eliminating the need for centralized identity providers. Role-based access control is facilitated through VC-Role, which defines the permissions of actors, and VC-Access, which ensures secure interactions with private blockchain channels.

The publication includes a description of the system architecture, DID and VC integration for access control, and a discussion of the QA requirements of the food industry.

The authors conclude that their system promotes traceability and ensures tamper-proof records of product quality. A proof of concept demonstrates the feasibility and potential impact of this approach in improving quality assurance.

This review (open access) covers food fraud in a wider context, including macro-economic motivation and opportunities for fraud.  The review has a particular focus on the Halal food sector within Islamic countries and communities.  After a discussion of modern “big data” analytical methods, such as spectroscopy and sequencing, it goes on to discuss point-of-use and real time testing approaches, sensors and internet-of-things, and predictive modelling  It concludes with the challenges in scaling some of these approaches, including inter-operability and data sharing, and makes a number of recommendations for capacity building in this field.  The authors propose a systems-level roadmap to bridge scientific innovation with regulatory and industrial application.

Photo by NASA on Unsplash

This study (purchase required) assessed the nutrient composition and labelling accuracy of twenty-nine commercially available insect-based pet foods: twenty-four dog foods and five cat foods.  All were labelled as complete and balanced. Twenty were labelled as hypoallergenic. The products were analysed for proximate composition, essential amino acids, and mineral content (calcium, phosphorus, potassium, magnesium, copper, iron, zinc, selenium, mercury, and molybdenum) according to AOAC guidelines. The ‘hypoallergenic’ products were assessed for animal DNA using next-generation sequencing.

The results were compared with label declarations, considering nutritional and legal tolerances, as well as recommendations from FEDIAF and NRC for the intended species and life stages (g/1000 kcal ME). Heavy metals were compared to maximum tolerable limits from the FDA.

The analysis revealed that 22 products (76%) did not comply with declared nutritional values and tolerances for at least one nutrient, with nine products (31%) showing discrepancies in two or more; key issues were in crude fibre and metabolizable energy. Three products (10%) met FEDIAF’s recommendations, and seventeen (59%) met NRC’s recommendations. Only one (3%) adhered to both label and FEDIAF’s recommendations. Most nutritional inadequacies were seen in selenium, calcium, phosphorus, Ca/P ratios, and taurine, potentially posing health risks to pets.

Fifteen out of twenty (75%) hypoallergenic-labelled products complied with the labelled species.

Despite the potential benefits of insect-based pet foods, this study underscores the need for further research and stricter quality control to ensure safety and efficacy, ultimately improving pet nutrition and consumers’ trust.

Photo by Hulki Okan Tabak on Unsplash

10 years ago, on the 14th July 2015, FAN was born - Happy 10th Anniversary to us! 🎉😍🏅🎂

Spectroscopic techniques are non-destructive, rapid, and often cost-effective tools for detecting cheese adulteration. Cheese is one of the foods most frequently reported as adulterated or misrepresented, particularly when including misrepresentation of PGO or PDI production methods or origin.

This review (open access) references 104 studies and describes the range of vibrational, nuclear magnetic, and mass spectrometric techniques which have been applied for cheese authentication, including Near-Infrared (NIR), Mid-Infrared (MIR), Fourier-Transform Infrared (FTIR), Raman, and Nuclear Magnetic Resonance (NMR) spectroscopy MS-based methods . Emerging non-invasive sensor-based technologies such as electronic nose (E-nose) systems have also been explored in dairy product monitoring and are covered in the review.

The authors consider that each technique offers distinct advantages based on its operational principle and application context. NIR spectroscopy, for example, has demonstrated utility in detecting water addition, milk source substitution, and fat adulteration in a variety of cheese matrices with minimal sample preparation FTIR and ATR-FTIR are valuable for functional group detection and surface compositional analysis, offering rapid screening capabilities . Raman and its variants, such as Surface-Enhanced Raman Spectroscopy (SERS) and Spatially Offset Raman Spectroscopy (SORS), provide molecular vibrational fingerprints useful for identifying foreign substances and analyzing samples through packaging.  ¹H NMR spectroscopy has gained prominence due to its high-resolution metabolomic profiling capabilities and its ability to differentiate PDO cheeses from non-authentic counterparts based on lipid and aqueous phase biomarkers .

Advanced mass spectrometry-based techniques, including LC-MS/MS and MALDI-TOF-MS, have also been effectively utilized for the detection of protein-based adulterants and species-specific peptides in complex cheese matrices, enabling quantification at trace levels.  Isotope Ratio Mass Spectrometry (IRMS) and other isotope-based techniques have proven crucial in verifying geographical and botanical origin by assessing stable isotope compositions such as δ¹³C, δ¹⁵N, and δ³⁴S

The authors aim to provide stakeholders—including researchers, quality control laboratories, and regulatory agencies—with an informed perspective on the strengths and limitations of each technique, thereby supporting the development of more robust authentication frameworks within the dairy industry.

Photo by Andra C Taylor Jr on Unsplash

The UK National Food Crime Unit is asking food businesses to be alert to the fact
that adulterated saffron has been identified on the UK market.

Saffron sampling results have revealed adulteration through the addition of
synthetic colours in contravention of Regulation 1333/2008.

Issues have also been discovered around saffron not meeting the grade stated on packaging (Grades I, II or III) due to lower than required levels of colouring strength (expressed as Crocin).

Failed samples are the subject of ongoing enquiries in the UK and internationally.

ACTION RECOMMENDED
If you are purchasing saffron either to sell or to use within a further processed product, please be aware of the risk of adulteration. Consider the following advice:

1. Ensure that reputable suppliers are being used and are being reviewed in line with supplier approval policies.
2. Be conscious of product being supplied at a lower price than would be expected for the quality and grade of saffron being purchased. If you have any suspicions, ask your supplier for traceability documents including any analysis reports.

Read full alert.

In this study (purchase required) a machine vision system was used to capture the images of saffron samples at different safflower mixture proportions. Then three feature extraction algorithms - gray level co-occurrence matrix, gray-level run-length matrix, and Local Binary Pattern -  were applied to extract the textural features of data. Discriminant Analysis, Support Vector Machine, and Artificial Neural Network algorithms as supervised classification models were applied to classify datasets.

The models were applied for 3 class and 6 class datasets to explore classification ability. The best outcome for the 6-class dataset was with the Support Vector Machine model and with all features with an accuracy of 80 %. For 3 class datasets, Discriminant Analysis model had the best result with all features and with the accuracy of 97.78 %.

To explore the statistical importance of different features, two Minimum Redundancy Maximum Relevance and Chi-Square Test algorithms were applied. For the gray level co-occurrence matrix extracted features, Chi-Square Test algorithm with 10 features had the best accuracy with a test accuracy of 76.94 %.

The authors conclude that the proposed approach could be utilized in designing a system for checking saffron authenticity at a business-to-business point of sale..

The Authenticity Methodology Working Group (AMWG) evaluates scientific research developed within Defra’s Food Authenticity Research Programme, ensuring methods developed are robust and fit for purpose. AMWG also advises on the wider application of methods.

This paper provides a summary of the group’s work in 2024. Activities include:

• providing technical direction on research to develop tools and methods for authenticity testing throughout the lifecycle of projects
• peer-review of final reports
• the provision of technical advice supporting specific issues to inform the development of policy on food labelling, composition and standards.

Read AMWG 2024 Annual Summary.

Unauthorized GM (UGM) refers to those GM crops or products that have not received approval in a particular country..

DNA-based methods are preferred for testing of raw or partially processed food products.  More heavily processed food can be more of an analytical challenge; DNA methods are still applicable provided the quality of DNA enables amplification despite being hampered by the processing procedures or presence of inhibitors or due to complexity in composition in terms of ingredients.

This review (purchase required) focused on possible approaches for adhering to the regulatory requirements while verifying UGM in processed food products. Technical challenges and approaches for extracting purified DNA and necessary quality checks are described. The article covers amplification as well as sequencing-based methods which can be applied to check for presence of UGM ingredients amongst a complex mix of other ingredients. The authors discuss the comparative assessment of these methods in the context of regulatory testing requirements and availability of information of transgenic DNA.

Photo by Kseniya Nekrasova on Unsplash

This new book from CRC Press is available to purchase either as an e-book or as hard copy.

After a general historical introduction to the adulteration of spices, the book then includes a chapter on global standards and regulations regarding spice purity (both adulteration/authenticity, and contaminants such as mycotoxins and pesticide residues).  It then has separate chapters giving an overview of adulteration and authenticity testing for specific spices; black pepper, chilli, nutmeg, saffron, ginger and turmeric.  After sections on mycotoxins and pesticides, it ends by discussing supply-and-demand pressures and the sustainable and ethical sourcing of spices.

Photo by Calum Lewis on Unsplash

The EC Monthly Reports of Agri-Food Fraud Suspicions reports are a useful tool for estimating fraud incidents, signposted on FAN’s Reports page.  The April and May 2025 reports have been added and can be found here.

FAN produces rolling 3-month graphical analysis. 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 subjective, intended only to give a high-level overview. 

• Unlicenced trade in high-risk foods (which includes attempts at illegal import, along with production from unlicenced operator sites) continues to feature frequently
• Fraudulent or missing traceability paperwork (including Heath Certificates and analytical test certificates) continues to be prevalent
• The increase in suspected false DoP/PGI claims in May was largely attributed to fraudulent Proscecco
• Ingredient quantities and quality feature persistently, particularly meat and fish in manufactured food
• Of incidents of traditional "adulteration", edible oils (particularly olive oil) and honey feature regularly.  Vanilla, coffee, spices, jams and spirits make more sporadic appearance.

As with all incident collation reports, interpretation must be drawn with care.  This EC collation is drawn from the iRASSF system – these are not confirmed as fraud, and the root cause of each issue is usually not public.  There are important differences in the data sources, and thus the interpretation that can be drawn, of these data compared to other incident collations.  For example:

• JRC Monthly Food Fraud Summaries (which underpin the infographics produced monthly by FAN member Bruno Sechet) - these are unverified media reports, rather than official reports, but hugely valuable in giving an idea of which way the fraud winds are blowing
• Official reports (as collated from commercial databases such as Fera Horizonscan or Merieux Safety Hud, which underpin FAN's annual Most Adulterated Foods aggregation) - these are fewer in number and give a much more conservative estimate of fraud incidence, and may miss some aspects which have not been officially reported
• The Food Industry Intelligence Network (Fiin) free SME Hub.  This excellent new resource collates anonymised UK industry test results for the benefit of Small and Medium Enterprises in the food sector (registration and approval required to obtain login).
• Verified reports (where the root cause has been scrutinised and interpreted by a human analyst, for example the FoodChainID commercial database) - these are also few in number, less suitable for drawing overall trends, but give specific insight and information.

If looking at trends over time, you must also be wary of step-changes due to the introduction of new data sources.  For example, Turkey's public "name-and-shame" database of foods subject to local authority sanctions went online in January 2025 and has resulted in an apparant increase in incident reports from Turkey.

Are you an English-speaking FAN member who also speaks either Arabic or Mandarin Chinese?

We have had FAN's Annual Summary translated into various languages using a professional agency.  We would like to sense-check these translations before pubication.  Would you be able to help?  We are not looking for editorial or grammatical corrections - merely a quick check that there is no inadvertent offensive language or any obvious scientific misunderstandings.  It should only take 5 minutes.

If you are able to help then please e-mail secretary@foodauthenticity.global.  Thank you

There are a large number of both commercial and in-house-written digital tools that attempt to classify and predict food safety risks based upon historic records in the EU Rapid Alert Service in Food and Feeds (RASFF) database.  With all such tools, it is important to remember that RASFF records are not a representative sample of either tests or results, and were never intended as a source of trends; the purpose of RASFF is rather to share specific individual alerts which may require regulatory action  on a cross-border basis.  Official tests are highly targeted, and often informed by previous RASFF alerts, so more alerts about a specific issue drives more official tests which drives more alerts (i.e. a feedback mechanism).  Also, RASFF only records the “positive” results, so there is no denominator; no indication of the number of “negative” results or the % incidence of an issue.  And finally, RASFF only records issues with a food safety concern so most food authenticity test results are excluded.

Despite these caveats, RASFF is still one of the most extensive and systematic public databases of food safety incidents and is likely to form the basis of many AI risk-prediction systems for years to come.

This paper (purchase required) evaluated the effectiveness of the Machine Learning models that sit behind such systems. The authors report that transformer-based models significantly outperform traditional machine learning methods, with RoBERTa achieving the highest classification accuracy. SHAP analysis highlights key hazards salmonella, aflatoxins, listeria and sulphites as primary factors in serious risk classification, while procedural attributes like certification status and temperature control are less impactful.

They conclude that despite improvements in accuracy, computational efficiency and scalability remain challenges for real-world deployment of AI risk-scoring and prediction systems.

Photo by Clarisse Croset on Unsplash

Nontargeted analysis for food authenticity by liquid chromatography–mass spectrometry (LC-MS) can provide data on thousands of chemical features. However, most studies that train machine learning models for food authentication use sample sizes in the tens or hundreds.  Such training sets are typically considered too small to be optimal, as it introduces the problem of overfitting when working with such a large feature-to-sample ratio.

This study (open access) aimed to mitigate this issue with a machine learning protocol designed for sub-optimal training sets, using honey as an example.   A recursive feature elimination (RFE) pipeline was developed specifically to address the challenges of optimizing the honey chemical fingerprint for multiclass machine learning classifiers on a limited number of samples with imperfect labels. A support vector machine was used for both RFE and classification to reduce the 2028 nontargeted features down to just 54 features (a 97.3% reduction) without any loss of classification performance.

The authors report that the resulting model was a 6-class classifier, capable of identifying monofloral blueberry, buckwheat, clover, goldenrod, linden, or other honey with a nested cross-validation Matthews correlation coefficient (MCC) of 0.803 ± 0.046. The development of a k-nearest neighbours filter and the decision to continue the RFE process beyond the iteration with the highest classification score were instrumental in achieving this outcome.

They conclude that this work shows a complete pipeline that automates feature selection from nontargeted LC-MS spectra when working with a limited number of samples and imperfect labels. This process can also be expanded to other food groups and spectral data.

Photo by Andrea De Santis on Unsplash