Enhancing Emergency Readiness in the Performing Arts with Free AI Tools

Weather hits. Power fails. Performers, staff and audiences get sick. When that happens, a simple plan saves time and money. Performing arts organizations are vulnerable to a variety of natural and human-caused disasters that can threaten their sustainability through harm to people, financial loss and damage to assets. Many have not developed emergency plans due to lack of capacity or uncertainty about how to start. New artificial intelligence-powered tools, such as Large Language Models (LLMs), can be great planning assistants. AI is not a replacement for professional advice, so think of it as a drafting partner. AI-generated drafts are a catalyst for conversation.

This is the first of two articles on using AI for emergency planning. This article will be an introduction for beginners who may only have access to free AI accounts. The second article will be for users with access to paid accounts and will describe the use of more advanced features. This first article looks at privacy considerations and usage limits that come with using free accounts. The field of AI is changing rapidly, and the content here reflects the technology as of November 2025. These tools will be very different a year from now.

Emergency planning is a process of gathering and analyzing information. LLMs, such as the ChatGPT or Gemini models, excel at these kinds of tasks. During emergency planning, an organization should perform a Risk Assessment that examines in detail areas of risk and develop things like Crisis Communications Plans, Cybersecurity Plans and Business Continuity Plans. Today’s LLMs can assist you with understanding what these are, why they are important and how to assemble them.

If you have opened a free account with OpenAI’s ChatGPT, Anthropic’s Claude AI, Google’s Gemini, or Perplexity, you have access to a tool with vast understanding of emergency planning; however, you should always check all outputs carefully and verify all citations as you would with any first draft. LLMs can “hallucinate” and give inaccurate information or cite fake references. One quick way to check a longer document is to have a different model review it, for instance having Gemini review what Claude has produced (this is not a substitute for human review). Always ensure any draft generated by AI is reviewed and approved by your leadership or board before inclusion in official plans. The second article in this series will address strategies to reduce inaccuracies.

Be careful about what information you give an LLM. Use caution and do not upload sensitive information such as door codes, passwords, social security numbers. When using a free account, remember that your prompts may be stored by the provider and could be reviewed to improve the model. We can create valuable planning documents using only public information about our organizations. Share only publicly available or general information in a free AI account. Details that appear on your website, or are already public, pose little security risk.

The documents you develop with these tools are just the beginning of planning. Your organization may want to create a committee or hire a consultant to further develop your plans. Be sure to include your local first responders in the planning process and find out if there are any emergency networks in your area. For deeper discussion of these topics, see the recordings of webinars on emergency preparedness for performing arts organizations on the Performing Arts Readiness (PAR) website. These webinars go into more depth about different aspects of planning. Additionally, sample emergency plans from performing arts organizations are available on the PAR website.

When prompting an LLM, attributing a role in your prompt can help the model focus on the relevant field. As an example for this article, we will use: # Act as an expert in emergency planning for performing arts organizations. Each line beginning with # represents a separate instruction the AI will interpret as a step or condition. (You should adapt any example prompts for your needs.)

When working with a free account, you can include detailed information about your organization or describe it more generally to give context. Such as: # Use verifiable, publicly available information about the Fake Not Real Theater, a concert venue in Atlanta, GA (Website: www.FakeNotRealTheater.---). Or, you could be more general and say # Consider a 500-seat concert venue in the Southeastern U.S. with a 50-year-old building with structural issues.

Clearly describe the task the model should do, for example: # Create a detailed business continuity plan outline tailored specifically for the Fake Not Real Theater. Include backup options for venue, staff, communications, and finances that fit its likely operations and scale…

Also, describe what you want in the response (format, length, content): # For each section, provide: ## A short explanation of why the section matters. ## Organization-specific recommendations using details about the Fake Not Real Theater. ## Example language or action steps written as if this plan were being drafted for that venue…

Now we can put the parts together and add more context for a complete prompt for the LLM:

# Act as an expert in emergency planning for performing arts organizations.
# Use verifiable, publicly available information about the Fake Not Real Theater, a concert venue in Atlanta, GA (Website: www.FakeNotRealTheater.xxx). If you cannot confirm specific facts, clearly state assumptions and label them as such. Do not invent or guess details.
# Based on what is publicly known or can be reasonably inferred from similar organizations, describe the theater’s key characteristics relevant to emergency planning (building type, capacity, staff size, location risks, and audience profile).
# Create a detailed business continuity plan outline tailored specifically for the Fake Not Real Theater. Include backup options for venue, staff, communications, and finances that fit its likely operations and scale.
# For each section, provide: ## A short explanation of why the section matters. ## Organization-specific recommendations using details about the Fake Not Real Theater. ## Example language or action steps written as if this plan were being drafted for that venue.
# Structure the response with clear headings for each section of the plan (Risk Assessment, Crisis Communication, Evacuation & Life Safety, Continuity of Operations, IT/Cybersecurity, and Recovery).
# Verify all details as accurately as possible, cite known facts when available, and clearly mark any assumptions made for illustrative purposes.

# Act as an expert in emergency planning for performing arts organizations.
# Consider a small regional theatre company with 10 full-time staff and 10 volunteers.
# Create a simple, step-by-step template for an emergency preparedness and response plan that can be filled out.

# Consider the needs of staff, performers, and audience members of all abilities.
# Include short explanations of why each section matters and how it can be adapted for limited staff and resources.

# Act as an expert in crisis communication for performing arts organizations.
# Consider a touring dance troupe with 10 performers and 3 administrative staff.
# Explain the main communication issues the troupe should plan for during a crisis, such as injuries, cancellations, or emergencies on tour.
# Create a fill-in-the-blank template for that troupe’s Crisis Communication Plan. ## Include sample crisis messaging.

# Act as an expert in emergency response and life safety for performing arts organizations.
# Consider an organization that manages a historic theatre building that seats 400 people in the Southeastern U.S.
# Create a template for an evacuation and audience-safety plan suitable for performances.
# Include checklists for staff roles, audience communication, and accessibility considerations. ## Consider ADA compliance and accessibility concerns.

You can also include any specific concerns that you have in your prompt such as artist cancellation, equipment failure or injuries: # Also consider the following risks specific to my organization: [list 2-3 key concerns here]. It can help identify accessibility gaps as well: # Review this plan for accessibility concerns.

Once you have a response, you can continue to refine the draft and drill deeper into topics as needed or expand the scope of the content with subsequent prompts. While remembering that the LLM is an assistant, not an authority, ask for explanations of things you don’t understand:

Provide more detail about the insurance policy needs mentioned in your response.
Ask me a series of Risk Assessment questions based on the information you have about my organization that I can use as the foundation of a risk assessment report.
Based on what you know about my organization, act as a facilitator of a tabletop exercise and guide me through a scenario where flooding has impacted my facility.

Most free subscriptions to the tools include limited use of a “Deep Research,” “Extended Thinking” or “Research” mode where the models spend more time conducting research and produce a longer, more detailed report. You will be limited to a few uses a month but try them out with a prompt you have developed.

Using an LLM to gather information and assemble a report is just the first step. You can follow up with these next steps:

Use the information you developed to work with a planning committee at your organizations.
Cultivate relationships with local first responders and share your plans with them.
Contact nearby organizations and include them in your planning.
Find out if there are emergency networks in your area that you can join.
If you are working with a consultant, show them what you have developed.
Consider the needs of your community as organizations with facilities can be tremendous resources during an emergency.
Update regularly and track versions (annually, after an incident or following organizational changes).

I hope this article has helped you start or revive your organization’s emergency planning!

These AI-powered tools are developing and will have new capabilities. The second article in this series will cover such things as Custom Instructions, CustomGPTs, and adding documents to a knowledge base. The second article will be published in early December 2025.