- 1. November 2026
- Posted by: AG CommTech
- Category: NEWS
Crisis Communication with AI: Speed Is Key, Trust Is Decisive
This article was written by: Christian Albrecht, Nuremberg Airport.
A crisis doesn’t follow a schedule. When a storm knocked down trees, damaged security fences, and temporarily brought flight operations to a standstill at Nuremberg Airport a few weeks ago, questions immediately arose: Can planes take off? What does this mean for travel plans? When will there be an update?
Situations like these show what AI in crisis communication is really all about. It’s not just about issuing quick statements, but about recognizing signals, organizing information, and acting based on verified facts. Technology can do a lot in this regard. But it cannot take responsibility.
First understand the crisis management process, then select the software
In the pre-crisis phase, risks and warning signs must be identified and procedures put in place. In the acute phase, it is crucial to assess the situation, coordinate decisions, and disseminate information through all relevant channels. This is followed by an evaluation: What information was available? Where did the reporting chain break down? Which decisions were effective?
There are AI applications for all three phases. But if you start by selecting tools, you’re starting at the wrong end. First, you need to clarify responsibilities, approvals, data sources, and communication channels. AI cannot fix flawed processes; in the worst case, it can actually exacerbate their weaknesses.
This is especially true for early detection. Social listening solutions such as Brandwatch, Meltwater, Talkwalker, or NewsWhip can highlight notable discussions, rising trends, and recurring signals. When monitoring data is combined with generative AI, it results in condensed situational overviews. An automated workflow can forward these directly to the responsible individuals in teams or via email. This creates a time advantage.
When the Boeing 737 MAX was grounded, it wasn’t until journalists asked about affected aircraft and passengers in Nuremberg that I realized the local relevance of this international event. A systematic, AI-supported analysis can help organizations relate global developments to their own operations and stakeholders at an earlier stage.
A core set of facts instead of five contradictory messages
During an acute crisis, information rarely arrives in a well-organized manner. Imagine a fire in a factory: A video of smoke is circulating on Instagram. Employees report seeing fire trucks. People on social media are speculating about an explosion. At the same time, reports are coming in from the control center, inquiries from the newsroom, and updates from plant management. What is confirmed, what is mere speculation, and what is still unknown?
This is where generative AI comes in: It can cluster incoming information, identify open questions, synthesize media coverage, and prepare drafts for situation reports. However, an unconfirmed social media post must never be turned into a supposed fact through linguistically persuasive wording. The organization needs a verified set of core facts in which confirmed information and uncertainties are clearly separated.
Based on this core set of facts, various formats can be developed: a holding statement as an initial confirmation, a Q&A for the press office, an employee bulletin, a website notice, or a social media post. This helps you move quickly while maintaining consistency. Those who rely on AI without first establishing effective communication channels confuse fluent language with reliable communication.
“Anyone who relies on AI without first establishing effective reporting channels is confusing fluent speech with reliable communication.”
This issue also affects other channels. At Nuremberg Airport, we follow a “website-first” strategy: In the event of an incident, the website serves as the central source of information to which other channels refer. A chatbot must therefore also be aware of the latest alerts. Otherwise, the website will state that operations are limited, while the chatbot continues to claim that everything is running normally. During our research prior to launching our chatbot, we observed such inconsistencies at other airports during a strike. Anyone digitizing communication channels must ensure that their information feeds are integrated with one another.
From deepfakes to a homemade action board
AI doesn’t just speed up effective communication. It also facilitates the spread of misinformation. An airline was confronted with an AI-generated video showing an alleged plane crash. The media also picked up on it initially. Google Lens, InVID-WeVerify, and proofs of origin help with verification, but they are no substitute for cross-checking sources.
On the production side, a communications team today doesn’t have to buy every application off the shelf. With Vibe-Coding, you can quickly build prototypes, even if you don’t have any programming skills. This allows you to test and gain experience. Hosting, data protection, and security must be addressed in the process.
AI is also a hassle-free sparring partner for training. A language model can take on the role of a tenacious journalist, asking critical questions one after another, challenging evasive answers, and ultimately providing feedback on factual accuracy, empathy, and action orientation. It is no substitute for a crisis management exercise, but it offers straightforward training.
Automate without outsourcing responsibility
Where is the line drawn? Aggregating data, extracting questions, translating reports, and drafting documents are tasks well-suited for AI. When it comes to making statements and providing recommendations to stakeholders, collaboration is essential. Humans must verify facts, assess risks, assign responsibility, or grant final approval. Genuine judgment and empathy are particularly crucial when it comes to potential victims, apologies, and expressions of sympathy. An organization cannot later claim that the AI phrased something incorrectly.
That is why binding rules are needed: Which systems may be used? What data may be entered? How are results verified and approvals documented? These rules should not stop at the door of the communications department. A crisis management team must also know how to use AI responsibly. And once the immediate crisis has passed, the same technology can help analyze logs, identify vulnerabilities, and improve processes.
The key question, therefore, is not how many AI tools a press office has. What matters is whether, in a crisis, it can quickly obtain verified information, use it to develop a consistent message, and clearly assign responsibility. AI can save valuable minutes. Trust continues to be built through clarity, reliability, and human decision-making.
Four hours of hands-on training with Christian Albrecht at the CommTech Academy.
Register nowChristian Albrecht on social listening, core facts, and the limits of automation.
Read the interview
Interview with Christian Albrecht: “AI Can’t Solve the Crisis”
AI agents need more than just good prompts
