Imagine this scene
A story has broken. Your phone is ringing. The communication director needs something they can say to the press in the next 20 minutes while the full response is being prepared. You need a holding statement — and you need it now.
In crisis communications, speed and accuracy pull against each other. The public and media want a response fast. Leadership wants to make sure the response is right. AI can help you move faster on the drafting side — producing structural scaffolding, standard holding language, and multiple tone options quickly — so that your human time is spent on accuracy and judgment, not on blank-page paralysis.
Key Insight
AI is useful in a crisis for speed on structure — not for judgment on substance. Never let AI decide the stance. That is always a human call, often a legal and senior leadership call.
A holding statement acknowledges a situation without committing to facts that haven’t been confirmed. It buys time. It signals that the organization is aware and taking the matter seriously. It does not speculate, assign blame, or make promises. AI can scaffold a holding statement quickly because the structure is predictable: acknowledgment, empathy or concern where appropriate, statement of action being taken, and next communication timing. What AI cannot provide is the factual accuracy or the decision about tone — those are always yours.
There are broadly three types of crisis drafts AI can help with: the holding statement (early response before facts are confirmed), the full public statement (once facts are known), and the internal all-hands message (parallel track for employees). AI can produce a starting draft of all three rapidly. Where it breaks down: when the situation involves legal exposure, when there is genuine harm to people, and when the facts are still unverified. In those cases, AI should be producing structural options only — every word must be human-verified before publication.
The safest workflow in a crisis: use AI to produce structural options and placeholder language quickly, then have a human strip out anything unverified and replace it with confirmed facts. Flag every claim in the AI draft for verification. Do not publish anything AI generated without a review pass, even in a fast-moving situation — especially in a fast-moving situation. The cost of a wrong statement in a crisis is almost always higher than the cost of a short delay.
Use AI to draft crisis materials for a realistic scenario, then identify every judgment call that would require human oversight before publication.
Choose a realistic crisis scenario: a product recall, a data breach, an employee misconduct allegation, a service outage, or an environmental incident. Write a brief description of the situation with only the facts that are confirmed so far.
Ask AI to draft a holding statement for this scenario. Review it and identify: anything that makes a factual claim not in your brief, anything that implies fault or liability, and anything that makes a promise you can’t verify will be kept.
Now ask AI to draft a full public statement for 24 hours later, assuming more facts are now confirmed (add 2–3 new confirmed facts to your brief). Compare the holding statement to the full statement structure.
Ask AI to draft a parallel internal message for employees covering the same situation from an employee-first perspective. Note how it differs from the external statement — and whether the differences are appropriate.
Highlight every sentence across all three drafts that would require legal review, leadership approval, or fact-verification before it could be published. This is your human oversight list — the work AI cannot replace.
