Brand crises in 2025 moved faster than traditional response playbooks could handle, with viral moments reaching tens of millions of viewers within 24-48 hours. The outcomes, however, defied simple categorization. Some companies that appeared to stumble in the initial response ultimately declared victory months later. Others contained damage through swift action. The one constant: AI search now archives every chapter of these stories, and whichever narrative has more authoritative content shapes what users encounter when they ask questions months or years later.

ChatGPT now reaches 800 million weekly users. Forty-four percent of consumers use AI-powered search as their primary information source. These platforms don't rank links; they synthesize narratives. When multiple narratives compete, the content with more authority, structure, and volume often determines what AI surfaces.

Three Brand Crises That Defined 2025

American Eagle: A Tale of Two Narratives

American Eagle launched an ad featuring Sydney Sweeney with the tagline "Great Jeans" on July 23. Social media mentions increased 18-fold. Brand sentiment dropped from +50 to -31 according to Signal AI data. Store traffic fell 3.9% in the first week, then 9% in the second week. The campaign generated 3,000 articles reaching 50 million readers, many of them critical.

By September, a different story emerged. CEO Jay Schottenstein told investors the marketing campaigns had driven "unprecedented new customer acquisition." The company reported 700,000 new customers and 44 billion ad impressions from the Sweeney and Travis Kelce partnerships. The stock, which had been down 20% from two years prior, recovered those losses and ended 2025 up nearly 45%.

What matters for reputation management: both narratives now exist in the information ecosystem. Someone using AI search for queries about American Eagle may receive either the crisis story or the turnaround story, depending on which sources the AI weights more heavily at that moment. Sources geared toward generative engine optimization (GEO) will tend to gain more traction.

Astronomer's Kiss-Cam Incident

At a Coldplay concert in July, a stadium kiss-cam captured Astronomer CEO Andy Byron with the company's chief people officer. The moment went viral on TikTok, accumulating 20 million views. Byron departed the company within days. Fake apology letters began circulating before the company issued its official response, forcing the crisis team to address fabricated content while managing the actual situation.

The Astronomer board acted within 72 hours on leadership changes. The swift action limited the window for speculation and prevented prolonged uncertainty about the company's direction.

Jet2's Viral Jingle

UK airline Jet2 released an ad with the slogan "Nothing beats a Jet2 holiday" set to an upbeat pop song. TikTok users began remixing the jingle with footage of travel mishaps: lost luggage, flight delays, airport frustrations. Over 1.3 million videos used the Jet2 sound. The company responded within 48 hours by joining the trend, launching its own TikTok challenge, and enlisting the original singer to participate. The initial mockery converted into brand engagement.

How AI Search Changes Crisis Narratives

Crisis coverage can get archived in AI training data or become a dominant feature when AI systems synthesize available information. But so can recovery coverage, recent earnings reports, and retrospective analysis. AI doesn't necessarily privilege the first story. It synthesizes from whatever authoritative content exists.

The American Eagle example illustrates this clearly. An AI system answering "What happened with American Eagle in 2025?" could emphasize the backlash narrative (sentiment collapse, traffic decline, social media criticism) or the turnaround narrative (stock recovery, customer acquisition). The AI synthesizes from what it finds, and multiple legitimate narratives may compete

This has direct implications for crisis response. The goal is not simply to survive the news cycle. The goal is to ensure that authoritative, well-structured content exists for whichever narrative you want AI systems to surface when users ask questions months or years later.

Generative Engine Optimization for Crisis Response

Generative Engine Optimization (GEO) is the practice of structuring and distributing content so it is more easily identified and cited by AI-powered search platforms.

When consumers ask platforms like ChatGPT, Perplexity, or Google's AI Overview about a brand, the answers are based on content these AI systems find during and after a crisis. GEO for crisis response involves working to ensure that content found by AI systems speaks to the narrative you want for your brand.  

What makes content citation-worthy for AI systems?

AI platforms weight sources differently than traditional search. Domain authority matters: content published on sites with established credibility (major publications, industry journals, official company pages with proper schema markup) earns more citations than equivalent content on low-authority domains. Semrush research indicates that companies with 50+ monthly mentions across authoritative sources appear in AI responses 320% more frequently than those with fewer than 10 mentions.

Structure determines extractability. AI systems favor content with clear headers, specific data points, and direct answers to anticipated questions. A well-structured FAQ addressing a crisis often becomes the source AI platforms cite months later. Content with quotations, statistics, and citations can boost visibility in AI answers by 30-40% compared to equivalent content without these elements.

Building content for multiple query types

American Eagle's situation demonstrates why crisis content strategy must anticipate different question framings. Create content that addresses:

• "What went wrong?" queries (acknowledge the criticism, provide context)
• "What was the outcome?" queries (document resolution, results, lessons learned)
• "Was it successful?" queries (present evidence, let readers draw conclusions)

Companies that only create defensive content cede the "outcome" and "success" queries to third parties. Companies that only trumpet recovery ignore the criticism that still appears in search results.

Structuring crisis response for AI extraction

Crisis response content should anticipate the questions AI users will ask. Use explicit Q&A formatting. Include specific dates, outcomes, and corrective actions. A crisis response page structured with clear headers and verifiable facts becomes extractable material that AI systems can cite when users ask what happened.

Combating misinformation in AI contexts

The Astronomer situation generated fake apology letters within days of the original incident. Misinformation compounds crisis damage when AI systems can’t distinguish fabricated content from official statements. Update your official website with a facts page. Post corrections on the platforms where false information spreads. 

Key Takeaways

• Multiple competing narratives can coexist in AI search; whichever has more authoritative content may dominate.
• AI search archives all chapters of a crisis story; create content for recovery queries, not just defensive responses.
• Misinformation monitoring is now an essential crisis infrastructure.
• Structured content (FAQs, comparison tables, specific data) earns more AI citations.

Frequently Asked Questions

How long do reputational crises last in AI search?

Unlike traditional search, where negative coverage can drop off the first page within months through SEO techniques, AI systems may continue surfacing crisis narratives indefinitely if that coverage dominates the available source material.

Can you remove crisis coverage from AI search results?

Not directly. AI systems synthesize available information rather than ranking discrete pages. The only method to improve AI representation is to create more authoritative, citation-worthy content that provides AI systems with alternative material to draw from.

Should crisis response content acknowledge criticism or only emphasize positive outcomes?

Both. AI systems respond to different query types. Users asking "What went wrong with [brand]?" will receive different results than those asking "Did [brand] recover?" Create content that addresses criticism with context, documents outcomes with evidence, and provides clear answers to both types of questions.

How do you monitor for AI-specific reputation issues?

Query major AI platforms (ChatGPT, Perplexity, Claude, Google AI Overview) with the questions your stakeholders might ask—including critical framings. Document how your brand is described, which sources are cited, and whether misinformation appears in responses. Tools can automate this monitoring across platforms.