Prepare the public response
How can you organise a rapid, consistent and verifiable crisis response without delegating sensitive decisions to AI?
3500 € per session, excl. VAT
In short
Two days, comprising 14 hours excluding breaks, to prototype a framework linking alerts, governance, messages and crisis monitoring. Using an evolving fictional scenario, you use generative AI to prepare materials, then test their reliability during a simulation. You leave with an operational dossier to be validated before any deployment.
Why this training?
In a crisis, a premature statement can turn uncertain information into a public commitment. The absence of explicit responsibilities also slows the response and encourages contradictions across channels. Generative AI speeds up the preparation of variants and difficult questions, but may invent facts or downplay a situation. The framework therefore requires traceability, verification and human validation before release. Running case study: Asterive Transport, a fictional school transport organisation, learns that a video accuses a driver of dangerous behaviour. The facts remain incomplete; parents, employees and a journalist are simultaneously requesting a response. Transfer to the workplace: Have the prototype validated by those responsible for communications, operations and legal affairs; schedule a limited internal exercise, measure response times and correct discrepancies before extending the framework.
Objectives of this training
- Mapping audiences and information risks
- Defining activation thresholds and responsibilities
- Prototyping source-backed messages with AI
- Testing the validation chain under pressure
- Measuring discrepancies and preparing a pilot
Who is it for and what are the prerequisites?
Roles and tasks concerned
- Communication and media relations managers
- Crisis team leaders and spokespersons
- Risk or public affairs leads
Prior skills
- Have experience in institutional or sensitive communications.
- Be able to formulate and revise a request to generative AI.
- Know your organisation’s decision-making processes.
What you will work on
Assessing the event
Objective: Define the scope of the information crisis. Weak signals, severity, uncertainty and exposed audiences.
Organising activation
Objective: Assign responsibility for sensitive decisions. Thresholds, crisis team, deputising arrangements and escalation.
Safeguarding AI assistance
Objective: Establish controls for AI-assisted production. Confidentiality, authorised sources, hallucinations and human oversight.
Writing the first message
Objective: Respond without speculating. Empathy, confirmed facts, actions initiated and the next scheduled update.
Coordinating channels
Objective: Avoid public contradictions. Internal information, media, social media and version control.
Anticipating objections
Objective: Prepare the spokesperson. Difficult questions, limits on responses and obligations to verify.
Structuring monitoring
Objective: Link signals to decisions. Monitoring log, rumours, urgent requests and corrections.
Simulating the crisis
Objective: Test the prototype under constraints. Roles, timeline and injections of contradictory information.
Evaluating and correcting
Objective: Assess the quality of the framework objectively. Accuracy, consistency, response times and traceability.
Preparing the pilot
Objective: Define the conditions for adoption. Scope, owners, maintenance and criteria for progressing to the next stage.
What you take away
- Activation and responsibility matrix
- Register of facts, uncertainties and validations
- Multichannel messaging and questions-and-answers toolkit
- Simulation report with assessment grid
- Pilot plan and priority corrections
Method and assessment
- Short input sessions followed by immediate practice using the case study that runs throughout.
- Individual work followed by work in pairs, with participants reviewing each other’s outputs.
- Guided group review: what works, what needs correcting and what remains a human decision.
- Preparation: Prepare an anonymised approval workflow, without personal contact details. Check access to an authorised AI tool; the exercises use fictitious or anonymised cases only.
Observable criteria: Each activation threshold identifies a decision-maker and their deputy. ; Each public statement refers to a verified fact; unknowns remain explicit. ; Each message specifies its audience, channel, approver and next update. ; The simulation measures alerting, approval and correction times. ; Each critical deviation results in a correction, an owner and a new test..
Frequently asked questions
Can we work on a real crisis?
Only after anonymisation; no sensitive information may be entered into an unauthorised tool.
Can AI publish the responses?
Not within this framework: any dissemination requires approval by a designated person.
Does the dossier replace legal advice?
No. Notification obligations and sensitive wording must be checked by the relevant functions.
Will the framework be rolled out in two days?
No. The training produces a tested prototype and a pilot plan that requires approval.
Last updated: