Artificial intelligence is beginning to transform fire risk assessment. We examine the capabilities, limitations, and ethical considerations of AI in fire safety compliance.. The AI Revolution Reaches Fire Safety Artificial intelligence and machine learning are beginning to penetrate one of the most traditional sectors of the built environment: fire risk assessment. From automated document analysis to predictive risk modelling, AI tools are offering the promise of faster, more consistent, and more comprehensive fire safety compliance. But the application of AI to life safety decisions raises profound questions about accountability, competence, and the appropriate role of technology in protecting human life. Understanding both the capabilities and limitations of AI in fire risk assessment is essential for building owners, fire safety professionals, and regulators. Current AI Applications in Fire Safety Document Analysis AI excels at processing and analysing large volumes of documentation: Drawing analysis — automated identification of compartmentation lines, escape routes, and fire safety provisions from architectural drawings Compliance checking — comparing building specifications against regulatory requirements Historical record review — extracting relevant fire safety information from building logs, maintenance records, and previous assessments Building Regulations interpretation — mapping building characteristics to applicable regulatory requirements Risk Prediction Machine learning models trained on historical fire data: Fire occurrence prediction — identifying buildings with elevated fire risk based on building type, age, occupancy, and maintenance patterns Severity modelling — predicting likely fire consequences based on building characteristics Resource allocation — helping fire services prioritise inspection and enforcement activities Insurance risk modelling — refined premium calculations based on granular risk data Image and Video Analysis Deficiency detection — AI analysis of photographs identifying fire safety deficiencies (missing fire stopping, damaged fire doors, blocked escape routes) Thermal imaging analysis — automated detection of electrical hotspots and insulation defects CCTV monitoring — real time detection of fire safety violations (propped open fire doors, blocked exits, smoking in prohibited areas) The Limitations of AI in Fire Safety Despite impressive capabilities, AI has significant limitations in fire risk assessment: Contextual Understanding AI cannot understand the nuanced context of a building's use, management culture, and occupant behaviour A propped open fire door may indicate a systematic management failure or a temporary maintenance operation — the AI cannot distinguish Fire safety is fundamentally about human behaviour, which resists algorithmic prediction Edge Cases and Novel Situations AI models perform well on common scenarios but struggle with unusual buildings, innovative construction, or novel risks Post Grenfell, the fire safety landscape is rapidly evolving — AI models trained on pre reform data may not reflect current requirements New risks (lithium ion batteries, mass timber, modern insulation materials) may not be adequately represented in training data Accountability and Liability Who is responsible when an AI generated fire risk assessment fails to identify a critical risk? The Responsible Person's duty under the RRO cannot be delegated to an algorithm Professional indemnity insurance may not cover AI generated assessments Courts have not yet tested the legal status of AI fire risk assessments The Hybrid Model: AI Augmented Expert Assessment The most promising approach combines AI efficiency with human expertise: 1. Pre assessment data gathering — AI processes building documentation, identifies potential issues, and generates a preliminary risk profile 2. Targeted physical inspection — human assessor focuses on areas flagged by AI analysis, plus professional judgement areas 3. AI assisted reporting — automated generation of standardised assessment sections with expert review and contextualisation 4. Continuous monitoring — AI powered building management systems providing real time fire safety intelligence between formal assessments 5. Portfolio analytics — AI aggregating data across building portfolios to identify trends, common failures, and prioritise investment Ethical Considerations The use of AI in fire safety raises important ethical questions: Transparency — building owners and regulators should know when AI tools are used in assessments Bias — AI models may embed biases present in training data (e.g., underrepresenting certain building types or communities) Competence displacement — over reliance on AI could erode human expertise over time Data privacy — AI systems processing building and occupant data must comply with GDPR Commercial pressure — AI tools may be marketed as replacements for qualified assessors, undermining safety standards For techno