AI won’t step in and manage a crisis for you

Teams who know how to use AI in a crisis? They’ve got the edge.

We run realistic, high-stress training exercises for communications, legal, and operational leaders all over the world. After hundreds of simulations, one thing stands out clearly: there’s a huge gap between teams who treat AI as a magic fix and those who know how to use it properly.

In response centres across the globe, crisis teams are already turning to tools like ChatGPT, Microsoft Copilot, and custom internal large language models (LLMs) trained on their own procedures. Some of our clients are even building dedicated AI assistants that can read through entire crisis response plans in seconds. We know AI is the ultimate way to get more done, faster.

But when the pressure is really on - how well does AI actually hold up?

The advantage: When AI works with you

Used the right way, AI won’t make decisions for people, but it will speed up the decision-making process. Teams that build AI into crisis management see four big benefits:

1. Faster scenario planning: They use AI to map out different possible outcomes, helping them predict how everyone from journalists to campaign groups might react in the first 24 hours.
2. Better starting communications: Instead of staring at a blank screen, leaders receive solid first drafts of press releases, internal updates, and Q&As ready in minutes.
3. Faster sense-checking: When information is coming in thick and fast, AI helps sort through reports to find the facts that matter most and flag what needs attention.
4. Staying aligned: Custom tools trained on your own policies help keep everyone following agreed processes, rules, and escalation steps.

The risks: Where AI falls apart 

Without practice, those same tools can turn into a major risk. When the pressure rises and things feel chaotic, it’s easy to rush, and make mistakes.

In our simulations, we witness the same critical errors over and over:

• The copy‑paste mistake: In a hurry, people often drop raw AI output straight into official messages. Sometimes they even leave behind placeholders like [Insert name here] or numbers and details the AI made up.
• Worse still, using AI without checking creates hidden problems:
  • A contradiction with your own rules: Generic tools often suggest standard responses that go against your organisation’s legal obligations or agreed ways of working.
  • Using AI for crisis apologies risks copying existing statements, making your brand sound insincere and formulaic. Communications teams must reverse-check AI content against past industry responses to ensure every statement is original, authentic, and genuinely human.
  • Believing what isn’t true: Under stress, people often stop verifying AI‑generated summaries and treat everything they say as fact, even when the details are wrong.

AI won’t rescue a team that doesn’t already know how to respond in a crisis, and it won’t fix broken ways of making decisions. It simply makes what your team already does, for better or worse, more visible under pressure.

Planning to use AI when a real incident hits? The smartest thing you can do is try it out first, in a safe, controlled environment.

Download our resource: The business case for AI in crisis management