New database brings together 392 million food safety records to track changing risks across Europe
Last Updated : 30 September 2026Brussels, 30 September 2026 – Around 392 million results from decades of European food safety monitoring have been brought together in a publicly available database developed by researchers from the EU-funded HOLiFOOD project.
The CompreHensive European Food Safety database (CHEFS) provides researchers a large-scale resource for analysing food safety trends and exploring how artificial intelligence could help identify changing contamination risks. In the future, tools developed using these data could provide additional evidence to help authorities decide where closer monitoring or sampling may be needed.
The database is one of the outputs highlighted as HOLiFOOD draws to a close. The project has published four policy briefs which translate recommendations for European and national policymakers. Together, they call for responsible use of AI, clearer pathways for emerging detection technologies, greater attention to how citizens perceive new risks, and policy decisions that consider impacts across the wider food system.
“Better data and new technologies can strengthen food safety, but they are only part of the answer. We also need expert interpretation, effective communication with citizens and policy decisions that consider consequences across the whole food system,” says Jeanne-Marie Membré, senior research scientist at Inrae and one of HOLiFOOD’s project partners.
Turning monitoring data into foresight
Food safety risks can be affected by climate and weather, global trade, socioeconomic conditions, and changes in food production. Understanding how these factors interact requires information from different parts of the food system.
HOLiFOOD researchers identified more than 300 variables that could be used as inputs for AI models, including environmental, socioeconomic, production and trade-related indicators. The project explored how these variables could be combined with historical monitoring results to identify patterns and support risk-based decision-making.
In one research application, historical data on mycotoxins, heavy metals, dioxins, and pesticides in animal feed were combined with weather and socioeconomic indicators. Researchers used the resulting dataset to develop an AI model that estimated the probability of a feed sample exceeding predefined safety thresholds.
This type of model is not an operational warning system and does not replace laboratory testing or expert assessment. However, it demonstrates how different sources of evidence could be combined to investigate where contamination risks may be changing and where additional surveillance could be considered.
HOLiFOOD’s recommendations emphasise that AI tools used in food safety must be transparent, explainable and built on well-documented and representative data. Researchers, authorities and other users need to understand how predictions are produced, assess their limitations and retain responsibility for interpreting the results.
New detection methods need clearer routes into practice
HOLiFOOD also investigated technologies that could complement established food safety controls, including metagenomics, portable PCR systems, biosensors, spectroscopy and AI-supported chemical analysis.
These technologies show potential, but wider implementation will require further validation and clear regulatory pathways. HOLiFOOD therefore calls for harmonised validation frameworks, guidance for AI-supported analytical methods, and interoperable data infrastructures that enable information to be compared and shared across countries.
Recognising a risk does not always lead to action
HOLiFOOD research indicates that risks associated with long-term drivers such as climate change may feel psychologically distant, meaning that people may recognise a threat without understanding how it could affect them or what they can do about it.
The project therefore recommends communicating emerging risks through locally and personally relevant examples. Citizen observations and social-media analysis may also provide information about emerging events and public concerns, but they have important limitations. Citizen-generated reports need to be verified and connected to institutional surveillance and response systems, while social media discussions should not be treated as representative of the wider population.
Food safety decisions have wider consequences
The fourth policy brief addresses the trade-offs that can arise when food safety, nutrition, environmental sustainability, food security, and affordability are considered separately.
For example, more intensive thermal processing may reduce microbiological risks while increasing energy consumption or contributing to the formation of process-related contaminants.
HOLiFOOD recommends involving policymakers, risk assessors and experts from different disciplines at an early stage, so that potential consequences across the food system can be identified and evaluated transparently.
ENDS
Notes to editors
About HOLiFOOD policy briefs:
The four HOLiFOOD policy briefs are:
- Holistic artificial intelligence for food safety
- Next-generation detection technologies for food safety
- Engaging citizens in emerging food safety risks: improving risk communication and identification
- Managing trade-offs in food systems: How holistic risk assessment can support better policy decisions
The briefs are available at https://holifoodproject.eu/policy-brief/
About HOLiFOOD
HOLiFOOD is an EU-funded research project bringing together expertise in hazard detection, artificial intelligence, data analysis, risk assessment, social science and citizen engagement. Its work explores how emerging food safety risks can be identified, assessed, communicated and managed.
For more information, visit https://holifoodproject.eu/
About the CHEFS database
The CompreHensive European Food Safety database consolidates approximately 392 million analytical results from official EU food safety monitoring. Developed within HOLiFOOD, it provides a publicly available resource for food safety trend analysis, risk-prediction research and the development of early-warning approaches.
The database is available at: https://github.com/WFSRDataScience/CHEFS
Media contact
For further comments or an interview with one of the project experts, please reach out to:
Lorena Vargas – EUFIC Media Manager
Email: lorena.vargas@eufic.org
