NAIROBI, Kenya – Artificial intelligence could soon play a critical role in protecting global food supplies by detecting toxic contamination before products reach the market, a new international study suggests.
The research, led by the University of South Australia (UniSA) and published in the journal Toxins, found that combining hyperspectral imaging (HSI) with machine learning (ML) makes it possible to rapidly spot harmful mycotoxins in food.
These poisonous substances, produced by fungi, contaminate crops like maize, cereals, and nuts during farming, harvesting, and storage.
Lead researcher and UniSA PhD candidate Ahasan Kabir said current methods are too slow, costly, and destructive to samples.
“In contrast, hyperspectral imaging captures detailed spectral information that allows us to quickly detect and quantify contamination across entire food samples without destroying them,” he explained.
Tests on cereal grains and nuts — highly susceptible to fungal contamination in humid conditions — showed the system could accurately distinguish contaminated from safe products.
The findings are especially significant for countries like Kenya, where aflatoxin outbreaks have repeatedly triggered food safety crises.
In 2004, contaminated maize killed 125 people and hospitalized more than 300 in Eastern Kenya.
The Food and Agriculture Organisation (FAO) has described maize contamination as one of Kenya’s most serious food security challenges.
Globally, FAO estimates that one in four food crops is affected by fungi that produce mycotoxins.
The World Health Organisation (WHO) warns that foodborne contamination causes about 600 million illnesses and 4.2 million deaths each year.
Researchers say integrating AI-driven HSI systems into food production and storage facilities could provide large-scale, real-time screening, preventing toxic foods from reaching consumers and strengthening global food safety systems.
