Handling a Brand Crisis When AI Assistants Repeat the Story

When a crisis hits, AI assistants can repeat what was published at the time, including old or incomplete information. You cannot delete those answers, but you can publish a clear, factual account of what happened and what changed, correct inaccurate sources, and track how answers evolve. Honesty works better than trying to bury the story.

First 48 hours: understand what is being said

Ask several assistants the questions people are likely to ask, for example “is [brand] safe to use” or “what happened with [brand]”. Note:

• Which facts are correct and which are wrong.
• Which sources are cited, where the assistant shows them.
• Whether competitors are now suggested instead of you.

Assistants that browse the web, such as ChatGPT with search, Perplexity and Copilot, may reflect recent news quickly. Answers based mainly on training data may lag behind and only change when models are updated.

Publish one clear source of truth

Create a dedicated page on your website that states, in plain language:

1. What happened, with dates.
2. Who was affected and how.
3. What you did to fix it.
4. What customers should do now, if anything.
5. Where to get help or more information.

Keep it factual and calm. Update it as the situation changes and show the date of each update. A clear page like this gives assistants and journalists an accurate source to draw from.

Correct inaccurate sources

If articles or listings contain errors, contact the publishers with documented corrections. Be polite and specific: quote the wrong sentence, give the correct fact, and link to evidence. Do not ask for accurate negative coverage to be removed. That rarely works and can create a second story. Our page on fixing AI errors about your brand covers this process.

Things that make it worse

• Flooding the web with thin positive articles to push the story down.
• Posting fake reviews or using fake accounts on forums.
• Denying facts that are well documented.
• Going silent, which leaves others to define the story.

These tactics can damage trust with customers and do little to change what AI tools say, since assistants weigh credible sources more heavily.

Weeks and months after: rebuild with real signals

Once the immediate issue is resolved, focus on genuine evidence of change: published safety updates, independent tests or audits where relevant, honest reviews from returning customers, and coverage of what you improved. Over time, these give assistants more recent, balanced material to draw on. See online reputation for AI tools for longer-term steps.

Monitor on a schedule

Set a fixed list of crisis-related prompts and check them weekly at first, then monthly. Track whether answers mention the resolution, whether outdated claims fade, and whether your official page is cited. Expect gradual change, not a sudden reset.

A note on legal and safety issues

If the crisis involves product safety, health or legal claims, involve your legal and regulatory advisers before publishing anything. Accuracy and required notices come first.

Prepare before a crisis happens

Keep a ready-to-use page template, a list of the publications and directories that matter in your industry, and a set of prompts you would check. Knowing who approves public statements also saves hours when time matters most.

How Keboost helps

Keboost helps you watch and respond. It tracks crisis-related prompts across ChatGPT, Gemini, Claude, Perplexity and Copilot, shows how you and competitors are described, and helps create and distribute clear, factual content so assistants can find your side of the story. Plans start at €29 per month. Run a free AI visibility check to see what AI tools are saying now.