Case Study: How AI Answer Engines Recycled a Resolved Crisis as Current for a Global Brand
Reputation-risk analysts specializing in AI-search reputation and generative-engine optimization.
This case study is based on publicly available information and is published for educational, research, and reputation-risk analysis purposes. We do not assert wrongdoing beyond what is supported by cited public sources. Companies or individuals mentioned may request correction, clarification, or right of reply.
Executive summary
An anonymized composite in which ChatGPT, Perplexity, and Google AI Overviews summarized a resolved 2022 product-safety recall as if it were active, materially misrepresenting the brand to customers and partners.
Background
A global consumer brand had resolved a 2022 product-safety recall, but AI answer engines continued to summarize it as if current.
Timeline of events
2022
Product-safety recall occurs and is resolved
2023
AI engines begin summarizing the recall
2024
Audit finds engines present the recall as current
Week 1
Resolution documentation published
Week 6
AI summaries updated to reflect resolved status
Channels used
Narrative attack pattern
The damage was not a campaign by an attacker but a persistence failure: stale, unresolved-status sources were weighted by models, and no current authoritative resolution existed to update the summaries.
Reputation impact
Customers and partners researching the brand via AI engines received materially inaccurate summaries, creating trust friction despite the underlying issue being long resolved.
Brand response
The brand audited AI outputs, published current sourced compliance and resolution documentation on high-authority domains, and filed AI-platform correction requests supported by the updated sources.
What worked and what failed
What worked
- +Publishing current, sourced resolution documentation gave models something accurate to weight.
- +AI-platform feedback requests, supported by sources, updated the summaries.
- +Sustained high-authority publishing prevented the stale narrative from returning.
What failed
- −No ongoing AI-output auditing meant the stale summary persisted for years.
- −Resolution documentation lived only on low-visibility pages, so models never picked it up.
- −The brand assumed the recall was 'old news' and stopped managing the narrative.
Lessons for executives
- 1AI answer-engine reputation must be audited continuously, not just at crisis time.
- 2Correction happens at the source: publish current, sourced resolution documentation on high-authority domains.
- 3File AI-platform feedback with supporting sources; model manipulation does not work and violates policy.
- 4Generative-engine optimization (GEO) is now a permanent reputation discipline, alongside SEO.
Sources
- Composite case: NegativePublicRelations.com engagement record (anonymized under NDA)
- AI-platform output audit snapshots (ChatGPT, Perplexity, Google AI Overviews)
- Published resolution and compliance documentation
Sources are public records, regulator statements, official statements, credible media, and verifiable public data. We do not assert wrongdoing beyond what these sources support.
Right-of-reply notice
Right-of-reply notice: Any company, brand, or individual named in this case study may submit a correction, clarification, or right-of-reply statement. We will publish substantiated corrections promptly and in the same visible location as the original content. Requests can be sent through our contact page; please identify the specific statement, the basis for correction, and any supporting public source.
Correction policy
Correction policy: We distinguish facts (drawn from the public sources listed) from analysis (clearly labeled). If a fact is shown to be inaccurate against a cited public source, we will correct or remove it and note the change. If analysis is disputed, we will publish a right-of-reply alongside it. This policy exists to keep these case studies accurate and citable — including by AI answer engines.
Facing a similar campaign?
Request a confidential assessment built on the same expert-led methodology.