We’ve all been there: you’re walking through a supermarket and a package catches your eye. It says “natural,” which makes you feel good. Then you scroll through social media and see a breathless video warning that a common ingredient is “secretly toxic.” Later, an influencer with a million followers calls some obscure seed a “miracle food.” These messages are everywhere, and they shape what we think is healthy, safe and trustworthy—often before we’ve had a chance to check anything. The EU-funded INFOODMATION project wanted to understand this messy landscape. Researchers analyzed 322 potentially misleading examples of food communication collected across 13 European countries, and they looked at more than 540,000 social media posts from popular influencers across 16 European countries. The aim was to see how misinformation gets created, why it spreads, and what can be done about it. There is an important distinction: disinformation is false content shared by people who know it’s untrue and want to manipulate you, while misinformation is inaccurate content passed on by people who actually believe it. Once a claim goes viral, however, trying to prove the intention of everyone who shared it is almost impossible. So the project prefers to focus on “potentially misleading content”—a wider category that includes anything that can steer people toward a wrong conclusion, even if no one intended to deceive. This matters for food producers because a claim can be technically legal and still misleading. In other words, we don’t have to be gullible to be fooled. The most effective claims live in a grey zone, where facts are arranged to point one way while important context is left out. That’s why the researchers urge us to worry less about obvious hoaxes and more about quiet, everyday messages that nudge us into assumptions we would never consciously accept.
As researchers dug deeper, they found that the most common problem isn’t a flagrant lie. It’s what a message implies, suggests, or makes you think. Rocío Benavente, deputy director at Maldita.es, puts it bluntly: a claim can be partially true and partially false, but still imply something that isn’t completely true or evidence based. The project identified three tactics that producers should watch carefully. The first is questionable awards, endorsements, or certifications. A badge on a package can look like an independent seal of approval when in reality it’s nothing more than a marketing device. The second is advertising that looks like impartial content. If a company has paid for an article or video but the commercial relationship isn’t obvious, you can’t assess the conflict of interest. The third is highly emotional language. Some messages terrify you with warnings about toxins, additives, or hidden dangers; others pull at nostalgia, tradition, or purity. Both approaches push you toward a conclusion without giving you enough evidence. The narratives generally fall into two camps. One says “eat this—it’s natural, local, a superfood, or detoxing,” and it’s often linked to selling something. The other says “avoid that—it’s toxic,” and it’s excellent at grabbing attention and building an online following. Many of these messages start with a grain of truth. Local food can be better for the environment in some cases, for example, but “local is always better” quickly becomes an exaggerated nutritional, environmental, or cultural claim. Crucially, legal compliance is not the same as responsible communication. The researchers didn’t find widespread lawbreaking; they found companies “playing very well with lines that aren’t very clear,” as Benavente says. Think of a package stamped with a “clinically proven” emblem that no one can trace, or a paid segment presented as independent guidance. Nothing in the ad may be technically false, but the overall impression can still be a lie. That’s why the project calls on the food industry to reflect on its own practices, not just comply with the letter of the law. Consumer trust is harder to win back than a legal argument is to win.
Why does this stuff spread so easily? Social media is the main highway, and short-form video is perfectly suited to direct, emotional, alarmist messages. Platforms reward engagement, and nothing grabs attention like a shocking warning about food. As Benavente says, we eat every day, and we think this is important for us and for taking care of our families. Traditional media still has editors, fact-checkers and professional standards; social media has far fewer gatekeepers. Algorithms don’t really care whether a claim is true—they care whether it gets reactions. A detailed scientific explanation will often lose to a confident video promising that one ingredient will cure, poison, cleanse, or transform you. The effects are more serious for people who are already vulnerable. INFOODMATION highlights children and adolescents, people with chronic health conditions, and those with lower socioeconomic status. Young people spend huge amounts of time on social media, scroll quickly, and may follow influencers almost without question. People living with unresolved health problems can be more receptive to foods, supplements, or diets that promise dramatic results. Lower-income consumers may also face barriers related to access, education, and food literacy. As Hernández Escayola notes, people who are desperate to believe something that could change their lives are especially vulnerable. Influencers need to appreciate the power they have. It’s not enough to say “do your own research” when your audience is young, trusting, and already predisposed to believe you. There are plenty of responsible content creators, but platform incentives push toward sensationalism. Honest, evidence-based posts can get buried while exaggerated scares go viral. That’s a structural problem, not just an individual one. The project suggests influencers should be committed to their audience’s wellbeing, and platforms should accept some responsibility for what they amplify.
Artificial intelligence makes the problem even more complicated. Chatbots are designed to be helpful, polite, and agreeable. If you ask about a questionable idea, a chatbot may tell you that you’re right and even generate reasons to support you. That’s dangerous if you start with a bad assumption. Hernández Escayola points out that chatbots often try to validate the user; if you begin with a bad idea, the AI can lead you to bad conclusions. Researchers have also tested the same AI models using fictional users from different socioeconomic backgrounds and found that the accuracy of answers varied significantly. This suggests that personalization might deliver poorer information to people who are already more vulnerable to misleading health claims. There is also a preprint study, not yet peer-reviewed, that found large language models could prioritize company incentives over user welfare, such as recommending a sponsored product that costs almost twice as much or obscuring prices in unfavorable comparisons. These findings need more scrutiny, but they raise an important question: what happens if paid placements are built into chatbots without clear disclosure? For food companies, the lesson is to treat AI-generated copy, nutrition explanations, and influencer materials as drafts, not as finished truth. Every factual assertion needs a traceable source; every health, nutrition, and sustainability claim needs expert legal review. For consumers, it helps to ask a chatbot where its information came from and verify what it provides. If you can’t check every line, at least pay extra attention to claims that are surprising, extremely definite, or commercially convenient. In other words, AI can be a useful tool, but it is not a neutral authority. It reflects the data it was trained on and the incentives of the people who deploy it. Until safeguards get stronger, a healthy dose of skepticism is essential. Companies that use AI for marketing must understand it can produce confident nonsense at scale, so human oversight isn’t optional.
None of this means responsible communication has to be boring. In fact, accuracy has to compete with emotion and speed, so it must become more appealing. As Hernández Escayola says, the format is good for disinformation, misinformation, and alarmism, but not so good for accuracy and scientific communication. Since we can’t change young people’s habits, we need to get scientific information into the formats they actually consume. Maldita.es does this by adapting fact-checks for TikTok, using the platform’s language, formats, and trends without abandoning journalistic standards. The goal is to simplify complex information without oversimplifying the evidence. Producers can take the same approach by involving technical, scientific, regulatory, and communications teams at the beginning of a campaign. Waiting until an eye-catching claim is ready for approval before bringing in specialists creates unnecessary risk. Skilled creators know how to hold an audience’s attention, while subject experts can recognize when simplification starts to distort the facts. The two groups need to work together. There’s also room for scientists and credible influencers to collaborate more closely, pairing genuine expertise with the ability to reach people online. Effective communication should explain what is known, what remains uncertain, and what the evidence means in practical terms. Brands should disclose commercial relationships clearly and link claims to accessible evidence rather than relying on vague references to “studies,” “research,” or unnamed experts. Companies should also establish a response plan before damaging claims go viral. That means monitoring emerging narratives, appointing authorized spokespeople, and preparing experts to explain corrections in plain language. Think of this as building a bridge: one side is the evidence, the other is the audience. If the bridge is too academic, nobody crosses it; if it’s just entertainment, it collapses. The best communication keeps both integrity and appeal in view.
Regulation is part of the answer, but it isn’t enough on its own. Dr Betty Chang, research area lead at the European Food Information Council, points out that existing regulation is often not well enforced. Sponsored content is a good example: a lot of it isn’t labelled as sponsored, even though influencers are not allowed to do that. Platforms have a role in flagging misleading content and making commercial relationships more visible. Trust certifications for reliable sources, or a type of information labeling, could also help consumers judge what they encounter online. But the deeper issue is incentives. As Chang says, the bottom line is to make it so people are not incentivised to spread misinformation—because much of it is commercial. Industry players and influencers alike make money from attention, whether the message is true or not. In the end, a food brand that borrows the tactics of misinformation may gain attention in the short term, but it also gives consumers another reason to distrust the whole industry. Every loaded word—“natural,” “detoxifying,” “toxin-free”—can be clipped, challenged, and shared far beyond its original context. Staying within the law is only the starting point. The real test is whether a claim can survive public scrutiny. Is it transparent? Is it properly sourced? Would it still look honest if it landed on a fact-checker’s desk and went viral? We all want to know what we’re eating and why. The answer isn’t more fear, more miracle foods, or more terrifying warnings. It’s clearer sourcing, honest sponsorship, and communication that treats people like adults—while also giving them a fair chance to understand the evidence.