By Bruce Johnson and Robert Marshall, UL Solutions
The fire service has changed over time as new tools and systems have been introduced to help firefighters do their jobs more safely and effectively. Some of the most important changes did not come from equipment alone, but from systems that reshaped how information is shared and how decisions are made.
The introduction of the Incident Command System (ICS) marked a major shift in how emergency incidents were managed. ICS replaced informal, locally developed approaches to emergencies with a shared structure for leadership, information flow and decision-making. In the early 2000s, the federal government formally required the use of the ICS because of its ability to solve real coordination failures.
| READ NEXT: AI notetakers in the firehouse — Policy, public records and practical use
The adoption of analytical tools like ICS also compelled the fire service to ask important questions: Who should rely on these systems? How should they be trained? What are their limits? And how do they support human decision-making rather than replace it? Over time, guidance, training programs and standardization helped put guardrails in place, so these systems strengthened safety instead of creating new risks.
AI has the potential to be another tool of that magnitude. A recent survey conducted by the NFPA found that 95% of respondents believe that AI has a purpose in at least some day-to-day job functions in the skilled trades.
It is already influencing how firefighters and officers research information, develop training and improve service delivery to their communities. As AI’s influence grows, the challenge for fire service leaders is not to decide whether AI belongs in a firehouse, but to ensure its use is guided by the same safety principles, oversight and accountability as previous technologies that affect life safety and public trust.
AI is already part of fire service work
To understand why guidance around AI is necessary, it helps to recognize how widely it is already being used. Firefighters are already leveraging AI tools to help explain unfamiliar topics, summarize complex documents or prepare for promotional exams. Officers are using it to outline training programs, draft policies and prepare presentations. Inspectors and fire marshals may rely on it to organize code research or summarize code requirements by occupancy type or create inspection checklists.
In most cases, this use happens informally. It is rarely written into policy, discussed in leadership meetings or addressed in training plans. Even so, AI is already influencing real work products. That makes it part of the fire service workflow whether departments formally acknowledge it or not.
This matters because anything that influences a fire department or emergency response agency carries responsibility. When a tool is shaping how information is created or interpreted, it deserves the same level of attention as any piece of equipment or procedure that has been adopted.
Understanding what AI is, and what it is not
Most AI tools in use today rely on large language models, which generate responses by recognizing patterns in vast amounts of text. They are designed to produce answers that sound clear, confident and helpful, which often makes them feel intuitive and easy to use. However, these systems are not checking facts, confirming whether information is current or evaluating whether an answer applies to a specific situation. They do not understand codes, standards or local requirements in the way a fire service professional does.
This limitation becomes apparent through what is commonly called hallucination, where an AI tool generates information that sounds reasonable but is incomplete or incorrect. Because these responses are often well written and confident, errors can be difficult to spot, especially without subject matter expertise.
Why accuracy matters more in the fire service
In many fields, AI can be used with relatively low immediate risk. However, the fire service operates in a very different environment, where decisions are closely tied to safety and public trust and depend on information that is precise, validated and current. When codes and standards change, it is typically because expectations around safety have changed as well.
AI systems do not recognize that progression. When asked a question about a fire code, they may present requirements or practices that sound reasonable but are based on older information without any indication that newer standards exist or that earlier approaches are no longer considered safe.
Take battery energy storage systems for example. These systems are being installed more frequently in buildings, infrastructure and energy projects as the use of lithium‑ion batteries continues to grow. While they support important goals like grid stability and renewable energy, the fire risks associated with large battery installations are still being actively studied and better understood.
Over the past decade, NFPA-led research has documented dozens of large-scale battery energy storage fire incidents in the United States. Many of these incidents involved early system designs, and each one contributed new lessons about how these systems behave during failures. As a result, guidance and code requirements have changed quickly.
However, an AI tool answering a question about battery energy storage has little contextual awareness of that history. It can be difficult to tell whether its response is based on early assumptions or on the most current safety expectations. Yet, the difference between outdated guidance and current best practice can influence how firefighters are trained, how inspections are conducted, how pre‑incident plans are written and how emergencies are handled when something goes wrong. In situations like this, confident but incorrect guidance can be more dangerous than having no guidance at all.
Experience and judgment still matter
The introduction of AI into the fire service does not change a fundamental truth of the profession: Experience still matters. As is common in many fields, younger firefighters and officers are often more comfortable experimenting with new technology, having grown up with it as a routine part of daily life. That comfort can be an advantage, but it can also come with risk. Less real-world experience can make it harder to recognize when an AI‑generated response is incomplete, outdated or subtly misaligned with current practice or code intent.
Experience provides context that technology cannot replicate. It helps firefighters and officers recognize when an answer feels too simple for a complex problem or when a recommendation does not align with how fires actually behave or how codes are enforced in practice. That instinct is built over time through training, exposure and responsibility.
Without guidance, there is a real risk that AI outputs will be accepted at face value, especially by those still building subject-matter expertise. Fire service leaders do not need to discourage AI use generally, but they do need to set expectations around how it fits into professional decision making. Understanding when AI can assist, and when human judgment must lead, is becoming part of the leadership role.
AI is reshaping codes and compliance beyond the firehouse
The influence of AI also extends beyond individual departments. Code authorities and standards organizations are beginning to introduce AI tools that are intentionally constrained to vetted, authoritative information. These systems are designed to help users navigate complex requirements while maintaining control over source material and interpretation.
At the same time, AI is being adopted by third-party companies to assist with plan review, code interpretation, and product evaluation. These tools vary widely in quality and transparency. Differences in training data, assumptions and verification processes can lead to inconsistent conclusions, particularly when AI outputs are treated as determinations rather than aids.
When AI can produce documentation that appears authoritative without being grounded in certification or testing, it can blur the lines between fact and fiction. That reality places a new responsibility for fire service leaders to question not just the conclusions they receive, but how those conclusions were formed.
Why UL 3115 is relevant to the fire service
As AI begins to influence safety-critical work across many industries, efforts like UL 3115 offer a useful reference point for how those systems can be examined through a safety and risk lens. While it is not written specifically for the fire service, the Outline of Investigation for AI-enabled products reflects a broader effort to bring structure and accountability to how the safety of certain AI systems and products are evaluated.
The principles behind UL 3115 are familiar to the fire service. They focus on understanding how a system is intended to be used, where human oversight is required, how risks are identified and how unintended outcomes are managed. These are the same questions fire professionals have long asked when new systems begin affecting training, operations or compliance decisions.
UL 3115 also highlights data use and protection, an issue with clear relevance for the fire service. Fire departments routinely handle sensitive information, and responsible AI use requires understanding what data should and should not be shared with AI tools, particularly those outside the organization.
A clear takeaway for fire service leadership
If AI is already influencing how information is generated and trusted within the fire service, then leadership becomes the deciding factor in how it shapes outcomes. Like any system that affects training, planning, or response, AI’s value depends on how clearly its role is defined and how deliberately it is integrated into professional judgment and accountability.
For fire service leaders, this approach is familiar. The profession has always placed safety first when new tools begin to affect how decisions are made. Applying that same discipline to AI is not about slowing adoption, but about ensuring that emerging technology supports sound decision making, protects trust, and aligns with the fire service’s long-standing commitment to safety.
ABOUT THE AUTHORS
Bruce E. Johnson is a regulatory services manager in UL Solutions’ Codes and Regulatory Services Department and manages regulatory operations in the U.S. Prior to March 2019, he held the role of senior regulator engineer, a position he held since joining UL Solutions in April 2015. Johnson serves on several ICC and NFPA model code development committees, including the IFC Code Development Committee, NFPA 1: Fire Code Technical Committees and NFPA 800: Battery Safety Code. Prior to joining UL, Johnson served 30-plus years in the fire, emergency services and life safety arena beginning as a volunteer firefighter on Long Island, N.Y. He also served as a fire marshal and N.Y. state-certified code official, retiring in 2008. Johnson worked for the International Code Council (ICC) as vice president, Government Relations, focusing on Fire Service Activities before joining UL Solutions in 2015.
Robert Marshall is a senior regulatory engineer at UL Solutions in the Codes and Regulatory Services (CARS) department. He is responsible for UL Solutions regulatory efforts serving the building and fire safety sector along with interacting with the International Code Council (ICC) and NFPA on code development activities. With over 30 years of experience in emergency response, fire prevention and public safety leadership, Marshall is a seasoned fire service professional. He last served as deputy fire chief for the San Mateo Consolidated Fire Department, and he has held a broad range of operational and administrative roles — from firefighter to fire marshal and division chief — building a reputation for strategic thinking, collaborative leadership and technical expertise.
