Artificial intelligence is transforming the way companies create, validate and maintain food labelling information. It can make drafting faster, help compare alternative wordings and make it easier to work with multilingual label texts. In food regulatory affairs, however, speed is only valuable when the final label is also legally compliant, accurate and suitable for the intended product category and market.
It is tempting to think that an AI tool could generate a compliant food packaging label within seconds. In practice, food labelling is rarely that simple.
A compliant label needs more than the right wording and a checklist approach: it requires a thorough understanding of the applicable legislation, product composition, claims, allergens, nutrition information, font size requirements, and of how the final product will be presented to consumers.
Labelling requirements also depend on the product category, as certain products, such as fortified foods, alcohol beverages or food supplements, are subject to additional, category-specific regulatory requirements.
At first glance, an AI-generated labelling text may appear complete and compliant. However, ensuring compliance with the applicable legislation is a more complex task. The responsibility for the accuracy and compliance of food labelling information remains with the food business operator, and mistakes in labelling can have commercial, regulatory and consumer safety consequences.
This blog post explores how AI can support the food labelling process, where the main pitfalls lie and why professionals in food regulatory affairs continue to be essential when preparing labels for the market. The focus of this text is on food labelling in the EU, where companies need to consider both harmonised EU legislation and national requirements in the target markets.
Food labelling is a regulated communication tool
Food labels are more than just marketing materials: they are regulated consumer information tools, intended to help consumers make informed choices and use products safely. In the EU, mandatory food information must be clear, accurate, easy to understand and not misleading. For prepacked foods, this usually includes the name of the food, ingredient list, allergen information, nutrition declaration, net quantity, storage and use conditions, preservability, business operator details and country of origin, where needed.
Creating food labelling texts is not simply a matter of applying specific rules: it usually requires detailed regulatory interpretation and product-specific assessment. Is the product name sufficiently descriptive? Do the ingredients need quantitative indication? Are the allergens emphasised correctly? Is the nutrition declaration presented in the correct format? Are the mandatory texts legible on the final packaging? Are all required languages included for the target market? These are not matters of wording alone; they are compliance decisions that require regulatory expertise.
How AI can help human experts in food labelling work
AI tools can support certain stages of the label preparation process. For example, they can help with drafting preliminary wordings and translations, improving clarity, comparing language versions and creating checklists for further expert review.
When used carefully, AI can act as a sparring partner: it can help to generate ideas, structure information and highlight points that may need further attention. This can improve efficiency, especially when managing large amounts of product information or preparing first drafts for multilingual labels.
However, any AI output should always be reviewed critically. AI can sometimes produce text that looks convincing but is incomplete, outdated, too generic or legally incorrect for the intended product category and market.
Common pitfalls of relying too heavily on AI
Nutrition and health claims illustrate well why human regulatory expertise remains essential. EU legislation permits the use of these claims only under specific conditions. AI may suggest compliant-sounding wording, but it does not necessarily determine whether the claim is suitable for a specific food product, whether the product composition supports its use, or whether all accompanying legal requirements have been met. Botanical health claims are a good example of this complexity. Around 2 000 claims related to botanical substances are still on EU’s “on hold” list and may currently be used under certain conditions, but their use requires careful assessment of the exact wording, substance, conditions of use and possible national interpretations of the regulations.
Another risk of using AI is accidentally leaving out mandatory information. Depending on the product, the label may need specific warning statements, usage instructions, nutrition details or mandatory information related to the product category. In many cases, legislation or authority guidance does not provide ready-made wordings for these texts. Instead, they often require regulatory interpretation based on the product’s composition, intended use, target consumer group and packaging. AI may not “know” the full recipe, packaging format, target market, national requirements or how the product is to be used, unless this information is provided in detail, and even then, AI may not interpret it correctly.
Allergen declarations require the same level of careful assessment. Allergens must be identified and presented correctly, and errors may have a direct and serious impact on consumer safety. AI can help with flagging possible allergen terms, but it cannot replace a thorough review of the recipe, verification of translated texts and validation of packaging artwork against product specifications.
Presentation requirements also matter. Mandatory food information must be legible and visible, and minimum font size requirements apply. A text may appear compliant in a draft document but fail in the final artwork if the packaging is small, the contrast is poor or the placement of graphic elements interferes with the presentation of the mandatory information.
Why mistakes matter
Labelling errors can delay product launches, create relabelling costs and, in worst cases, lead to product withdrawals or recalls. Even when the error is not an immediate safety concern, making corrections to printed packaging is expensive and takes time.
Errors can also pose a risk to the reputation of the brand. Consumers and business partners, as well as authorities, expect food information to be reliable. Incorrect allergen information, misleading claims or missing mandatory particulars can weaken trust in a brand and unnecessarily increase the workload of customer service, quality assurance and regulatory teams.
Why a food regulatory affairs professional is still essential
A food regulatory affairs professional brings to the table the context that AI, by default, does not have. The human expert understands how general EU requirements interact with product-specific rules, national legislation, authority guidance and practical artwork limitations. They know how to assess whether mandatory information is complete, whether marketing text is in line with the regulations and whether mandatory labelling elements are correctly positioned on the packaging. They can also check if the language used is natural, accurate and appropriate for the target market.
Most importantly, a regulatory professional knows how to navigate the subtle details and nuances that often come up in food labelling. These small details can change the regulatory conclusion: the amount of a vitamin or mineral, the function of an ingredient, the target consumer group, the physical form of the product, the wording of the marketing materials or the country in which the product will be sold. Furthermore, experienced regulatory professionals continuously build practical knowledge that goes beyond what is stated in legislation and public guidance. They are familiar with authority expectations communicated during inspections, training sessions, industry events or discussions with competent authorities. This background knowledge is extremely important in food labelling, where applying the legislation in practice often requires interpretation and where public guidance does not cover every possible situation.
Best practice: combine AI efficiency with expert review
The most effective approach is not to reject AI, but to use it in the right way. When the underlying product data is accurate and complete, AI can help to generate drafts, organise information and improve consistency. This requires reliable information of the recipe, ingredient specifications, nutrient values, intended use, product category, target markets, packaging details and any claims or marketing messages. Final regulatory review and approval should nevertheless be carried out by a qualified food regulatory affairs professional who can assess the label against the exact product specification, composition, packaging format and applicable legal requirements.
All in all, AI is a valuable assistant in food labelling tasks, but it is not a substitute for regulatory expertise. Food labels must be legally compliant, truthful, clear and suitable for the intended market. When AI-generated content is combined with a professional regulatory review, companies can operate more efficiently while still protecting compliance, consumer safety and brand reputation.
Need support with food labelling? Medfiles’ Food Labelling Team can support you with the details and practical implementation of labelling requirements for foods and food supplements. We continuously monitor regulatory developments and provide up-to-date support for food companies of all sizes, from early label drafting to final regulatory review.
Author:
Maija Salmenhaara
Team leader, Labelling // Regulatory Affairs Expert, Food and feed
Maija has worked at Medfiles from 2019. Her expertise covers regulatory affairs in food and feed, and she has strong competence in regulations concerning food composition and safety, food labelling, nutrition and health claims, novel food ingredients, food supplements and fortified foods.
She is also very experienced with work related to animal feed composition, labelling and marketing claims, with a balanced set of skills in project management, client service and writing reports.
Maija has a M.Sc. degree in food sciences (nutrition) from 2007. Before Medfiles, Maija has worked both in governmental agencies and the private sector, focusing on nutrition research, risk assessment and market research.
Free webinar: Overview of food labelling in the EU
This webinar provides a general overview of voluntary and mandatory food labelling information as well as specific tips for different areas and current hot topics in the field. You will gain a concise information package of the different aspects of food labelling and the key regulations governing the field.
See also:
