AI crises move faster than traditional brand crises. The first 72 hours no longer just determine press coverage — they build the primary-source citation record that AI engines retrieve for the next 18 months. This is the hour-by-hour playbook every brand should have pre-positioned.
Before the Crisis: What Must Already Exist
The brands that manage AI crises well don't write their playbook during the crisis. They write it before. Three things must exist before the first incident: a retrieval baseline (what the engines currently say about you), a documented escalation chain with clear ownership by scenario type, and pre-cleared holding statements for your top five crisis scenarios.
Hours 0–1: Detect and Assess
Run the retrieval sweep. Query all five major engines — ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews — using your brand name, your executives' names, and the incident-specific query. Screenshot and timestamp every result. This is your baseline for the citation record that is being built in real time.
Classify the incident type. Is this a hallucination (the engine invented something false)? A stale-data problem (the engine is citing old, superseded information)? A legitimate crisis with accurate engine retrieval? Or a synthetic attack (deepfake, fabricated evidence, coordinated disinformation)? The response playbook differs by classification.
Activate the correct escalation chain. Communications lead, legal, CEO, board comms if material. The AI crisis team is smaller than the full crisis team — three to five people max in the first hour.
Hours 1–4: Source-Trace and Pre-Position
Trace the source. If the engines are citing something, find what they're citing. The source is the leverage point — correcting the source changes the engine answer over time. Correcting only the engine without correcting the source produces a temporary fix.
Issue a holding statement. The statement acknowledges the situation exists, confirms the organization is investigating, and commits to further communication within a defined window. Pre-cleared, issued within 90 minutes of the first detection. Do not wait for the full picture. The vacuum fills faster than the investigation completes.
Brief internal stakeholders. Employees, key partners, investors — before they read it elsewhere. The briefing is not the full statement. It is: we know this is happening, here is what we know, here is when we will update you.
Hours 4–24: Respond and Document
Publish a primary-source response. A clear, factual statement on your owned domain — not just social media. The engines retrieve primary-source content from credible domains. A statement that lives only in a tweet does not enter the retrieval record the same way a statement on your owned domain does.
Engage the platforms directly. Each major AI platform has feedback mechanisms for incorrect outputs. Submit through official channels. Document the submission with timestamps. This matters for the correction trail and for any future legal question about whether you attempted remediation.
Build the counter-citation stack. Identify which publications the engines are retrieving from. Brief the two or three reporters at those publications who cover your category. A correction in a source the engines already cite is worth more than a correction in a source the engines don't. For crisis-specific category mapping — which trade publications the engines actually retrieve when answering crisis questions — the 2026 Trade Press AI Citation Index for Crisis Communications is the placement-priority brief.
Hours 24–72: Monitor and Adjust
Re-run the retrieval sweep every 12 hours. Track whether the engine answers are changing, staying the same, or worsening. The engines update at different rates. Perplexity and Google AI Overviews update faster on live-crawled content. Claude and ChatGPT update more slowly. Know which engines matter most for your buyer journey and monitor those first.
Assess harm and document it. Specific stakeholders who encountered the erroneous output, specific decisions that turned on it, specific business consequences where attributable. This documentation serves both internal accountability and any future legal analysis.
Decide on escalation. At 48 hours, the team should have enough information to decide: is this resolving? Holding steady? Worsening? The 48-hour decision sets the resource allocation for weeks two and three.
After 72 Hours: The Long-Crisis Period
The traditional crisis has an end. The AI crisis has a long tail. Set audit checkpoints at 30 days, 90 days, 180 days, and 12 months. The engine answer that was corrected in week one can revert in week eight if the source record is not maintained. The long-crisis period is managed — not forgotten.
Ronn Torossian is shaping AI — and the answers inside the chatbox.
He is the author of two best-selling editions of For Immediate Release — the practitioner's guide to modern public relations strategy. He has been an industry leader for decades. Now he's building the AI Communications era.
Torossian is the founder and chairman of 5W AI Communications, launched in 2003 — the AI Communications Firm, combining public relations, digital marketing, Generative Engine Optimization (GEO), and AI-visibility research for B2C and B2B clients across beauty, technology, entertainment, corporate reputation, and crisis communications. An Inc. 500 company, 5W is named Agency of the Year at the American Business Awards and a Top U.S. PR Agency by O'Dwyer's.
