Survey on the use of AI in food labeling

IMERO regularly hosts webinars on food labeling—especially when products are marketed internationally and different labeling requirements apply. We conducted two short surveys in 2025 in two webinars with different food manufacturers. The participants were mainly quality managers and those responsible for quality assurance, regulatory affairs, and product management who are interested in the use of AI for international labels.

We asked whether and how AI is already being used today—from general tools to specialized solutions. The results are surprisingly clear: AI has already arrived in the everyday life of many companies, but is still much less prevalent in labeling itself.

1) Have you already implemented AI in your company?

Results

• Yes: 74%

• No: 26%

For many companies, AI is no longer just a topic for the future. The high “yes” rate shows that experimentation or productive work is already underway in many areas, such as text drafting, internal evaluations, process automation, and translation tasks.

Interestingly, most participants have already used AI. This suggests that there is a fundamental openness to it – and that internal teams are increasingly developing routines for integrating AI tools into processes in a meaningful way. At the same time, “AI used” does not automatically mean “AI established in critical compliance processes”: Especially where mandatory food labeling and regulatory details play a role, additional security and verification mechanisms are often expected.

---

2) Have you already used AI for labeling?

Results

• Yes: 32%

• No: 68%

Labeling is not simply a matter of “layout and text.” It is the interface between product data, design, translation, quality assurance, and regulatory requirements. In many companies, labels are subject to approval processes, internal responsibilities, and external audits.

Only a small proportion already use AI directly for labeling. An obvious reason for this is that the requirements are high because they involve labeling obligations and correct mandatory information on food products – in other words, content that can quickly come under scrutiny in everyday market life (retailers, authorities, consumers). In addition, the discussions revealed that many teams have not yet achieved the desired transparency in terms of legal and regulatory references with conventional AI solutions. This slows down implementation, even if there is great interest in general.

---

3) Which AI have you already used for labeling?

Results

• General LLMs (ChatGPT, DeepL, Gemini, etc.): 41%

• Specialized AI: 10%

• None: 49%

A typical pattern emerges here: General tools are readily available, easy to test, and deliver results quickly—especially for text-based tasks. Specialized solutions are used less frequently because there is less choice, implementation is usually more structured, and a clear definition of objectives is often necessary.

About half do not yet use AI for labeling. At the same time, 41% rely on common providers of general AI models – presumably because they can quickly assist with tasks such as text variants, wording aids, or initial translation of labels. The 10% share that already uses specialized AI is also interesting: this indicates that some companies are already looking very specifically for solutions that are tailored to certain processes.

This is precisely where specialized approaches typically come in: instead of “just generating text,” the focus is on comprehensible results, structured product data, and clearer derivation from applicable specifications. At IMERO, the focus is on supporting labeling processes through a combination of AI and expert knowledge—e.g., with functions for label checking, rule-based notifications, and updates when requirements change..

---

4) What did you use AI for in labeling?

Results

• Translation: 14%

• Label checking: 9%

• Research: 33%

• Not yet: 44%

It is particularly telling that 33% currently use AI primarily for research in this context. This shows that AI is often used first as a “sparring partner” to structure information, gather initial clues, or prepare questions for internal/external experts. At the same time, 44% say they do not yet work with AI in labeling. In the webinar discussions, this was often less a matter of rejection than caution: many teams first want to ensure that processes remain transparent and that work on labels is reliably documented and verifiable.

---

Between curiosity and responsibility

Our webinar surveys paint a picture that many quality managers will probably recognize: AI has arrived in companies—but when it comes to labels, the bar is set higher. And that's understandable. A label is not only a means of communication, but also a component of product conformity and thus a process in which teams would rather check once too often than once too little.

The bottom line is that the surveys paint a very practical picture: AI has basically arrived in many companies – but when it comes to labels, it is (still) used selectively. Where the benefits are quickly apparent, such as in research (33%), it is already being tried out. However, when it comes to more demanding steps such as label verification (9%), the requirements for traceability, process integration, and transparency increase. And the fact that 44% have not even started yet makes it clear that the market is in flux, but many decision-makers want to take a controlled and structured approach to getting started.

Today, there are effective ways to achieve this: clearly defined use cases, clean databases, and an approach that takes quality assurance and documentation into account. IMERO supports companies in making international food labeling more efficient—with AI-supported workflows and expert know-how—so that teams can save time without losing track of labeling requirements.

If you are currently at this point, a pragmatic next step could be to clearly define which tasks AI should support—e.g., translation of labels, structured preparation of text modules, or systematic label checking as an additional audit trail. It is crucial that results remain traceable and fit into existing approvals.

IMERO is developing a solution that brings together AI and expertise to help teams reduce the coordination effort involved in international food labeling, minimize risks, and move more quickly from research to implementation.

If you would like to discuss this further, we look forward to your questions—whether for a single target country or for an international rollout strategy.