McGill University researchers use AI to spot fake honey

Researchers at McGill University have developed an innovative technique based on artificial intelligence to determine the floral origin of honey. 

“We had already worked on honey to find antibiotic or pesticide residues,” explains Stéphane Bayen, director of the Department of Food Science and Agrochemistry at McGill University. “Over time, we realized that we could see thousands of other chemical substances in these samples. This is what led us in recent years to use this broader fingerprint to try to identify honey.”

The chemical fingerprint of the honey is then analyzed by algorithms which, using specific markers, make it possible to identify the origin of the product. 

This new protocol has the advantage of not only being faster—results are obtained in minutes rather than days—but also more accurate. Until now, the origin of honey was verified by pollen analysis, a less reliable method when the product is processed or filtered.

Unique markers for each type of honey

“What we call a marker will be a molecule that will be unique, or that will be higher, in a honey sample,” explains Professor Bayen.

Blueberry honey, which is very popular in Quebec, for example, is characterized by the presence of niacin. “We can find many other markers, but this one was very interesting because it’s still a vitamin. It speaks to consumers a little,” said Bayen.

Buckwheat honey, on the other hand, is distinguished by its concentration of phenolic compounds, known for their antioxidant properties.

Chemical fingerprinting offers much more than simple authentication. It also allows for the assessment of various aspects related to honey quality, such as its healthiness and product safety. “We can detect, in the same analysis, compounds such as Hydroxy-Methyl-Furfural (HMF), which will tell you if the honey is fresh or if it has been sitting on shelves for a long time, for example,” explains the professor.

Fight against fraudulent labeling

As the honey market grows, misleading labeling is becoming an increasingly concerning practice, as highlighted by the Canadian Food Inspection Agency.

“I’m careful with the word ‘fraud’ because it would be intentional,” warns Bayen. “A producer doesn’t know exactly where his bees are looking for nectar.” This doesn’t prevent the researchers, given the results obtained, from being uncertain about the good faith of some producers. “For 10 to 20 per cent of the samples, we had doubts about the labels,” he says. 

The botanical origin of honey influences not only its taste but also its nutritional value. This reality leads some to mislead consumers, resulting in unfair competition for beekeepers. For example, monofloral honey is generally sold at a higher price than an equivalent polyfloral honey.

“The honey industry and beekeepers are still an industry that receives little support and access to research and development, compared to other industries. It was a great pleasure for me to work with honey professionals because they have a lot to gain from being supported by science,” concludes Stéphane Bayen. 

Although this technique is still in the experimental stage, the research team hopes it will soon be adopted by food inspection agencies around the world. The researchers have already begun adapting it to other products, including beverages.

–This report by La Presse Canadienne was translated by CityNews