Misinformation your clients bring you
Clients will bring you rumors and screenshots. Learn why false claims travel further in your clients' languages, and how to correct them without shaming anyone.
Write the correction together
- Name one false claim your clients are hearing right now. Write it on the board exactly as they hear it.
- In pairs, write three sentences that correct it without making the person who believed it feel stupid.
- Read them out, take the lines that worked, and keep them as the team's script for that rumor.
Which correction would you be willing to say to your own family?
- False claims reach your clients with less friction. Fact-checking labels and reliable reference sources are sparser in many client languages, so a rumor that would get flagged in English can keep circulating untouched in Spanish, Arabic, or Dari.
- People who rely on social media in their own language for news can be more exposed to false narratives. This reflects the information environment around a community, and it says nothing about how careful any client is.
- AI can write fake posts that read as more convincing than human-written ones, and it can produce realistic fake images and screenshots. A polished, professional look is no proof that something is true.
- One AI tool can produce a flood of tailored false messages aimed at specific groups, so the same false claim can reach different communities dressed to fit each one.
- Correcting a false belief takes more than saying it is wrong. A claim can keep shaping how someone thinks even after they accept the correction, so give the accurate information plainly, explain briefly why the false version is off, and point to a source they trust.
- Warning people about a manipulation technique before they meet it builds resistance to it, and short in-language, culturally grounded correction can shift behavior. Lead with the client's concern and never make the person who brought the rumor feel foolish.
Fact-checking labels and reliable reference sources are thinner in many of your clients' languages, so a false claim meets less resistance before it reaches them. People who rely on Spanish-language social media were more likely to believe false political narratives (Abrajano and colleagues, 2024). The pattern reflects the information environment around a community, and it says nothing about how careful any client is. The stakes are real. Among Latinx patients hospitalized for COVID-19, misinformation and fear of immigration consequences combined to delay people from seeking care (Cervantes and colleagues, 2021).
AI now makes this harder. In one study, disinformation written by an AI system was more convincing to readers than similar messages written by people, and readers could not reliably tell which was which (Spitale and colleagues, 2023). A single public AI tool has also been shown to produce more than a hundred tailored health-disinformation posts aimed at different groups, along with realistic fake images, with little effort (Menz and colleagues, 2024). A screenshot that looks polished and official can be manufactured, so appearance carries no weight as proof.
Correcting a false claim takes care. Research on how people handle corrections shows that a false belief can keep influencing someone even after they accept that it was wrong (Ecker and colleagues, 2022). Telling a client they are wrong rarely settles it. Giving the accurate information plainly, saying briefly why the false version does not hold, and pointing to a source they trust works better.
Two approaches help. Warning people about a manipulation technique before they encounter it builds resistance to it, an approach called prebunking or inoculation (Roozenbeek and colleagues, 2022). And correction built in the client's language and grounded in their community can change behavior, as a culturally designed in-language vaccination campaign did for the community it was made for (Castellon-Lopez and colleagues, 2024).
Correcting without shaming
- Affirm the concern. Show the person it makes sense they paid attention to it.
- Give the accurate information in plain language, and say briefly why the false version is wrong.
- Offer a source they can trust in their own language, and write it down.
- End with a next step, so the correction turns into something they can act on.
Correcting without shaming, step by step
A client brings a screenshot that scared her
- A client shows you a screenshot forwarded in her family group chat, written in her language. It claims that using food assistance will get her green card application denied and her family reported. She is frightened and has already skipped an appointment because of it.
- You start with her feeling. You tell her it makes sense she is worried and that many people have seen messages like this one. You do not tell her the message is foolish or that she should have known better.
- You look at the screenshot together. You point out that it names no source, carries no date, and shows no official logo, and that AI and forwarding can make a message look official when it is not.
- You give her the accurate information you can stand behind, in plain language, and you are clear about what you can confirm yourself and what a benefits or legal specialist needs to check for her exact case.
- You offer a source she can trust in her language, such as a known legal-aid line or an official page in her language, and you write it down so she can share it back into the same group chat.
- You help her rebook the appointment she skipped, so the correction leads to a concrete next step for her.
Practice correcting a rumor gently
Role-play in pairs. One person shares a rumor they saw in their language. The other practices naming a trusted in-language source and offering the correction warmly, without making the first person feel foolish. Switch roles and debrief which phrases landed well.
Map the rumors you are already seeing
As a group, list the false claims your clients have actually brought to you, in whatever language you heard them. For each claim, write down the accurate information and one source in that language you would send a client to. Notice which languages have almost no trustworthy source you can point to, and flag those gaps for the team to fill. Keep the list where everyone on the team can reach it, and add to it as new rumors show up.