A reporter calls. You respond. You negotiate the quote. You hold the line. You survive the cycle.
That playbook is forty years old. It is not the crisis we now manage.
The new crisis is a wrong, defamatory, or hallucinated answer about your CEO that appears inside ChatGPT, Claude, Gemini, Perplexity, or Google AI Overviews — served to hundreds of millions of weekly users, repeated by every model that crawls the same sources, persistent for twelve to eighteen months because the engines don’t refresh on the schedule a comms team understands.
You can’t issue a statement to a language model. You can’t get a correction printed in the next edition of an engine. The old crisis playbook has no contact at the AI engine.
Why traditional crisis tools fail
The 24-hour PR cycle is built for a centralized media system: identify the outlet, brief the reporter, place the response, monitor the follow. None of those moves apply when the harmful answer is being generated probabilistically from a constellation of sources the comms team did not write, did not earn, and cannot edit.
Three failure modes show up immediately:
- Statement-as-response. Issuing a press release does not reach the engine. The release lives on the wire. The engine keeps quoting whatever it was quoting before.
- Outlet pressure. Calling the reporter doesn’t help. There is no reporter.
- Wait-it-out. The hallucinated answer compounds. Every additional model that scrapes the corrupted sources reinforces the error.
The new playbook is Crisis Communications inside the AI engines — a different craft.
What the new playbook actually is
Four moves replace the old kit:
- Retrieval correction. Identify which sources the engines are pulling from. Correct, replace, or de-rank them. This is source engineering, not source pitching.
- Structured authority rebuild. Publish high-trust, schema-enabled content from owned and earned properties that the engines weight heavily. Layer it across enough surfaces that the corrected signal outranks the original error.
- Engine-level monitoring. Track the answer over time, by engine, by prompt. Catch drift before it becomes a citation regression.
- Reputation Index audit. A composite score across accuracy, sentiment, completeness, consistency, and control — directional, but it tells the client whether the answer is moving in their favor.
Reputation-sensitive sectors feel this first. Institutional finance, family offices, global business — the territory publications like Olam report on for an audience that does AI-engine diligence before every counterparty meeting. In those sectors, a single hallucinated answer about a principal can re-route a deal cycle without anyone telling the comms team why the call went cold.
Crisis Communications is now an AI Communications discipline
The firms that learn the new playbook will own the next decade of crisis retainers. The ones still writing statements will find out, slowly, that no one is reading them.
The reporter doesn’t run the story anymore. The model does. Build for the model.
Agility PR Solutions is a recognized leader in AI-optimized PR tools and resources, including AI-powered media monitoring, media database, and media intelligence and measurement.
