Generative AI has made convincing fake images, video and audio cheaper and more accessible than ever. For crisis communication that creates a new risk: disinformation spreading faster than the facts you are trying to share. This article explains why that is, what the Dutch threat assessment says about it, which response models exist and how to prepare practically.
Why deepfakes are a communication risk, not only a technical one
It is tempting to see deepfakes as a technical problem: a detection question for IT specialists and fact-checkers. That view is too narrow. For anyone responsible for crisis communication, a deepfake is above all a communication problem with a technical cause. A manipulated video of a public official, a cloned voice in a phone call or a forged press release can take on a life of its own within hours, while the organisation itself is still establishing the facts.
The World Economic Forum places disinformation, substantially amplified by generative AI, firmly among the leading global risks for 2025. That is not an abstract observation. It means organisations must reckon with a crisis consisting not only of the incident itself, but also of a second, parallel information crisis: the question of what people believe they are seeing and hearing, and how fast an incorrect picture spreads before a correction catches up. Anyone still organising crisis communication purely around sharing their own verified information risks finding the narrative was long ago set by something that never happened.
The Dutch threat assessment
Dutch security services point this out as well. In the report "Amplified threats in a world full of artificial intelligence" (December 2024), the AIVD, MIVD and NCTV jointly conclude that generative AI lowers the barrier to creating and spreading disinformation and propaganda. That applies not only to state actors operating with geopolitical motives, but explicitly also to non-state actors. The report names groups such as Al Qaeda and ISIS as examples of actors able to deploy these capabilities.
For organisations in the Netherlands this means the threat is broader than "a hostile state targeting us". The accessibility of the technology means smaller, less predictable actors can also produce convincingly manipulated material, with potentially large impact on a crisis already under way or on one caused by the disinformation itself. That makes preparation relevant for practically any organisation involved in crisis communication, not only for the most obvious targets.
From reactive to proactive: what practice teaches
Research among PR professionals into AI deepfakes and strategic crisis communication shows practice is still largely reactive: acting only once a deepfake or disinformation campaign is already visible, with the risk that the organisation is permanently behind the facts. That same research identifies a clear need to move towards a more proactive approach, with concrete strategies: including deepfake scenarios as standard in the crisis communication plan, strengthening stakeholder relationships so trust is already in place before it is needed, improving media monitoring and sentiment analysis to catch early signals, and working structurally with fact-checkers and technology partners rather than seeking them out in the moment.
The conclusion is clear: a purely reactive approach, however understandable under time pressure, is less sustainable than structural preparation. That aligns with recent refinements of Coombs's Situational Crisis Communication Theory (SCCT), which since 2022 has been sharpened in successive publications specifically for misinformation crises. This refinement helps determine which response strategy (denial, diminishment, rebuilding or bolstering) best suits how a particular piece of disinformation is spreading: is it an isolated message, a viral wave, or a coordinated campaign? The right strategy depends on that spreading pattern, not on a standard response identical for every situation.
Practical steps to prepare
From these insights, a number of practical steps can be distilled that are achievable for most organisations, even without a large specialist team.
First, establish a protocol in advance for verifying images, video and audio. Who inside the organisation judges whether material is genuine, which steps are taken to do so, and within what time? A protocol invented during the crisis costs valuable time you do not have.
Second, appoint a fixed point of contact for possible deepfakes and disinformation. This prevents reports from staff, customers or media getting stuck between departments, and creates one place where signals come together and a decision is taken on the approach to follow.
Third, communicate transparently once something has been rebutted. Explain what exactly turned out to be false, how you established that, and what is true. This transparency strengthens the organisation's credibility over the longer term, even though it costs extra explanation in the moment.
Finally, apply the insight from the refined SCCT: choose a response strategy fitting how the disinformation is actually spreading, rather than denying by default or responding at length by default. An isolated message with little reach calls for a different approach than a video going viral fast. Thinking that distinction through in advance means the response fits the situation better at the moment it is needed.
More on countering disinformation during a crisis
CrisisRadar publishes an extensive guide on countering disinformation during a crisis, with concrete handles for verification, response and collaboration. In Dutch.
Sources
- AIVD, MIVD and NCTV: Amplified threats in a world full of artificial intelligence (December 2024), in Dutch.
- World Economic Forum: The financial impact of disinformation on corporations.
- Wiley: research into AI deepfakes and strategic crisis communication.
- Situational Crisis Communication Theory (SCCT): Refining and Clarifying a Cognitive-Based Theory of Crisis Communication.
- ScienceDirect: refinement of SCCT for misinformation crises.
About this article
Published by CrisisRadar B.V.
This article has been checked against primary sources: Dutch crisis management practice, European regulation and academic literature. The sources used are listed above, with dates. Spotted something that is wrong? Tell us; we will correct it and record the change. Read how we work on Editorial policy.
Try it yourself with CrisisMaatje
CrisisMaatje connects the AI you already use — Claude, ChatGPT or Copilot — to the CrisisRadar platform. You ask for help in plain language, your AI drafts a message and records the facts, the timeline and the statutory reporting deadlines in a crisis file. You review it and decide what goes out. Free, no credit card.