Researchers at McGill University have developed a cutting-edge method that uses artificial intelligence to authenticate the origins of honey, tackling a longstanding issue of fraud in the global food trade. This novel approach, which relies on advanced chemical analysis and machine learning, ensures that the information printed on honey labels accurately reflects the contents of the jar. For years, food scientists and regulators have grappled with widespread mislabelling in the honey industry, and this breakthrough could offer a practical, scalable solution to a problem that has proven difficult to address with conventional techniques.
Dr Stéphane Bayen, Associate Professor and Chair of McGill’s Department of Food Science and Agricultural Chemistry, emphasised the importance of this work: “Honey is among the most fraud-prone commodities in global trade,” he noted. “A great deal of the fraud arises from misrepresenting either the floral source of the honey or the region in which it was produced.” Monofloral honey derived predominantly from a single flower type, such as acacia or manuka, are particularly susceptible to fraud, as they are highly prized for their distinct flavour profiles and associated health benefits. Their rarity and appeal allow them to command significantly higher market prices, creating a strong incentive for unscrupulous producers to mislabel their products.
What makes the McGill method especially significant is its ability to identify the floral sources of honey, even in highly processed samples where traditional authentication methods fall short. Until now, verifying a honey’s origin typically involved pollen analysis, requiring intact pollen grains. This approach, however, becomes unreliable when honey is filtered, pasteurised, or otherwise refined. Instead, the new technique uses high-resolution mass spectrometry to generate a comprehensive molecular fingerprint of the honey. This fingerprint is then interpreted by machine learning algorithms, which compare it against reference profiles from known floral and geographical sources.
Testing has shown that this AI-based technique is accurate and far faster than existing methods. In trials, the researchers successfully classified a range of honey samples according to their botanical origin, including those that had been processed and would have posed challenges for pollen-based identification. “Whereas traditional authentication could take days, and often fails with filtered honey, our system delivers results within minutes,” said Dr Bayen. This increased efficiency opens the door for more widespread and routine testing, making it easier to catch fraudulent products before they reach the market.
Beyond ensuring consumer transparency, the researchers highlight their work’s broader ethical and economic implications. The pressure on producers to compete has increased with the growing demand for local and artisanal food products—such as Quebec’s blueberry honey. This technology could serve as a protective mechanism for honest beekeepers who risk being undercut by fraudulent competitors. At the same time, consumers are increasingly concerned with sustainability and food integrity, and tools like this can provide much-needed reassurance that they are purchasing genuine products. As Bayen remarked, “People deserve to know that their honey is what it claims to be, and honest producers deserve protection.”
The McGill team sees considerable potential for applying this technique beyond honey. Many high-value food products—such as olive oil, wine, saffron, and even coffee—face similar issues of mislabelling and adulteration. Integrating AI with chemical fingerprinting could thus form the foundation of a broader shift in how food authenticity is verified. The researchers are now seeking partnerships with food safety agencies and industry bodies to see their technique adopted as a standard tool in quality assurance and fraud prevention across the food supply chain. Combining rigorous science with technological innovation offers a promising solution to an increasingly pressing global concern.
More information: Stéphane Bayen et al, Rapid Convolutional Algorithm for the Discovery of Blueberry Honey Authenticity Markers via Nontargeted LC-MS Analysis, Analytical Chemistry. DOI: 10.1021/acs.analchem.4c01778
Journal information: Analytical Chemistry Provided by McGill University