A recent publication from the Food and Agriculture Organization of the United Nations (FAO) has provided an in-depth review of early warning systems for food safety risks, an explanation of available open access tools, and the potential applications of Big Data and artificial intelligence (AI) in the field.

Early warning systems play a crucial role in the reduction of food safety risks from various hazards, and they enable swift implementation of mitigation measures when issues arise. Recent developments in early warning reflect a shift away from reactive systems toward proactive systems that are based on predicting emerging food safety risks. Modern warning systems require real-time and diverse data, and can be enhanced by AI and machine learning (ML) technology.

Different methods and systems can be used for the timely prediction and identification of food safety hazards and their public health implications, such as early warning and response systems (EWARSs) and foresight. Such methods may focus on different timeframes (i.e., present, near-term, or long-term), hazards, and health effects.

In recent years, significant progress has been made with Big Data and AI applications in food safety early warning and emerging risk identification—such as biosensor, Internet of Things (IoT) and Blockchain technologies—as well as ML techniques like Bayesian and neural networks. However, gaps and barriers exist to the adoption of these tools for food safety early warning and emerging risk identification, especially in low- and middle-income countries (LMICs), where data reliable internet access and sufficient infrastructure and facilities for data collection, storage, and processing may be challenging. From a socioeconomic perspective, better coordination of food safety activities at national and sub-national levels, building partnerships among different stakeholders, and capacity development through the provision of sufficient training for all food safety actors are important.

Aside from technical and socioeconomic barriers, the implementation of AI historically comes with a number of ethical and policy challenges. Concerns include protecting privacy and surveillance when personal data are used, the unintentional inclusion of bias and discrimination into the design of tools, safeguarding ethical decision-making from wrongly influenced human judgement, and the consideration of intellectual property (IP) rights. The efficiency of decision-making informed by AI has also been questioned, as the centralized infrastructure of AI-based decision models may lead to delayed response time at sampling sites.

To assist authorities, especially in LIMCs, the report also highlights and presents a range of tools and methods for food safety early warning and emerging risk identification, and provides detailed practical information for professionals on three open access tools—MedISys, MedISys-FF and SGS DIGICOMPLY—covering their functionalities and usage.