Managing a crisis with AI tools demands more than fast prompts: it requires verification, cultural readiness and clear protocols. If you lead communications, marketing or operations, knowing how to filter noise from real threats can protect your brand reputation when fake news, hostile comments or deepfakes hit your channels.

How do you use AI prompts during a real crisis?

When you decide to lean on artificial intelligence to navigate a crisis, the quality of your output depends on the quality of your input. Think carefully about how you write each prompt, what context you provide and, above all, how you will review every detail afterward.

The recommendations an AI gives you are only useful if they match the real situation you are managing. Read twice. Question the answer. Adjust before acting.

What should you check before trusting an AI response in a crisis? Verify that the advice fits your real context, confirm the data sources and review every recommendation against what your team knows on the ground.

Are all negative comments a real crisis on social media?

Short answer: no. You may suddenly see hundreds of negative comments on your social channels, attacks in reviews of your restaurant, your delivery app or your business. Before you react, filter the noise and verify.

Not every wave of criticism is organic. Some accounts are not real, some campaigns are coordinated, and some narratives are amplified by algorithms. Even so, the impact can still reach you, so you need to be prepared either way.

How do fake news and algorithms amplify a crisis?

Fake news is another form of attack: false stories that elevate the conversation and force you to respond. Many crises have escalated because the company did not react in time with a clear narrative, a statement or a publication.

Algorithms add fuel. We have seen cases where high volumes of comments tried to influence public perception. A useful reference here is the Cambridge Analytica story, an older case that still leaves valuable lessons about how data and volume can shape opinion.

Do all negative comments need a public response? No. Sometimes communication must be direct and private. Other times, strategic silence is the right move while you gather facts.

When is strategic silence part of a crisis strategy?

Silence can be a deliberate choice, but only when it is strategic. It is not the same as taking too long to respond. The Louvre Museum robbery is a clear example: the delay in releasing a first statement opened space for rumors and false stories to spread.

You also need to assume hostile actors are watching. They will use every news cycle, every mistake and every opportunity to damage your reputation. That is why response time and message control matter so much.

A simple traffic light helps you classify each situation:

• Red code: high risk, immediate action required.
• Orange code: medium level, monitor and prepare a response.
• Green code: under control, keep listening.

How do you handle a deepfake attack on your company?

Imagine you see a video of a well known executive from a large corporation saying things that make no sense. You are likely facing a deepfake: a video manipulated with artificial intelligence so it looks like the real person is speaking, when they are not.

These attacks are growing. Picture this: a video leaks showing someone inside your food company claiming the ingredients are low quality and cause illness. Comments explode online. How would you manage it?

How do you protect your team from AI scams and deepfakes?

Deepfakes of executives giving instructions are increasingly used for scams. Suppose your team receives an urgent money request from your CEO through a video call where they appear briefly and then hang up. Would the person on the other side question it?

The answer depends on the culture and protocols you have built. Work on three layers:

• Internal communication that empowers people to question unusual requests.
• Verification codes or steps to confirm that an instruction is real.
• Action protocols so no one acts on a suspicious video or message without checking first.

Trust inside your team is what allows someone to pause and say this does not add up before transferring money or sharing sensitive information.

What is a deepfake? It is a video, audio or image manipulated with artificial intelligence to impersonate a real person, often used in scams or reputation attacks.

Which tools help you detect AI manipulated content?

There are tools available today that help you identify whether a video or a design has been intervened with artificial intelligence. Use them as part of your verification routine, especially when content goes viral or arrives through unusual channels.

Artificial intelligence can support you during a crisis, and it can also be the weapon used against you. Building cultural awareness, clear protocols and verification habits is what turns your team into the first line of defense.

How would you handle a leaked deepfake about your company? Leave your answer in the comments.