Artificial intelligence has begun finding its way into many aspects of public safety operations, including the drafting of fire and EMS reports. Departments are experimenting with software that can summarize run sheets, correct spelling and grammar, and even generate narrative text based on prompts. As the technology becomes more accessible, questions are emerging about how AI fits within the legal and evidentiary role of incident documentation.
At the heart of the issue is a basic function of any fire or EMS report: to capture the contemporaneous observations of responders and document the actions they personally took at a point in time when the details are fresh in their mind. Reports serve as the written record of what a firefighter, EMT or paramedic saw, heard and did at the time of the incident.
Our legal system has rules that treat firsthand reports written by factual observers favorably in light of their inherent reliability. These rules include the business records exception to the hearsay rule, the past recollection recorded rule and the use of records to refresh a witness’ recollection.
AI-oriented questions
The business records exception allows properly created fire and EMS reports to be admitted into evidence, because they are made at or near the time of the event by someone who has firsthand knowledge as part of the agency’s regular operations.
The past recollection recorded rule allows a report to be read into evidence when the responder no longer remembers key incident details, provided that the report was accurate when it was made.
When a testifying responder does remember the event but needs help recalling specifics, a report can be used to refresh a witness’s recollection, allowing the responder to review their own contemporaneous writing before testifying.
When the text of a report is not written by the individual who made the observations or took the actions, the reliability of the documentation is inevitably called into question. A report written by someone other than the firsthand observer/responder—whether another crew member, an administrative assistant or an AI program—risks disconnecting the narrative from the perceptions and memory of the person who actually witnessed the event. If AI generates portions of that narrative, questions will inevitably be raised about whether the report accurately reflects the responder’s own recollection or reflects a reconstruction of the events produced by an algorithm. This can become problematic when the report becomes evidence in a criminal proceeding, civil litigation, disciplinary action or administrative review.
AI-oriented benefits
AI tools can play a productive role when used as an adjunct to traditional report writing. Many responders struggle with spelling, grammar, sentence structure or efficiently organizing information. AI can assist by ensuring that key factual elements are not omitted, that narratives flow logically and that reports meet an agency’s formatting expectations. Some systems can even prompt the user for missing information—such as the patient’s level of consciousness, timeline of events or actions taken—helping to ensure that the final report is complete and factual.
AI can also enhance readability. Clear, well-organized reports reduce misunderstandings and make it easier for investigators, attorneys and medical directors to follow the sequence of events. Spelling and grammar corrections further prevent confusion, without altering the underlying facts.
However, these benefits depend on the responder maintaining control over the content. The facts must originate from the individual who was on scene. The responder must verify that the final narrative accurately reflects their own observations. AI should serve as a tool—much like spellcheck or a documentation checklist—rather than a substitute author.
Firsthand knowledge
As AI becomes more integrated into fire and EMS documentation software, fire departments may encounter new questions about authorship, accuracy, authentication and the evidentiary trustworthiness of incident reports. However, the central principle remains unchanged: A report must reflect the firsthand knowledge of the responder who signs it. Within that boundary, AI-assisted tools can help strengthen clarity, completeness and consistency, while preserving the integrity of the responder’s account.
