Home>Fire>Risk Assessment>Putting AI to work
ARTICLE

Putting AI to work

19 November 2026

ARTIFICIAL INTELLIGENCE is already present in the UK’s fire safety sector, for the most part thanks to tools no-one in positions of leadership approved. Unsupervised Artificial Intelligence can pose a real risk, but approved and responsible Artificial Intelligence is transformative for efficiency, accuracy, fewer errors, less time spent on administration tasks and more attention being focused on transacting the vital work that keeps people safe. Jasper Bartlett delves into the fine detail.

Artificial Intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence. Those tasks include learning, problem-solving, reasoning, perception and, what’s more, decision making.

As far as the fire safety sector’s concerned, make no mistake. AI is already here, whether you’ve planned for it or not. In California, a network in excess of 1,200 surveillance cameras watches the hillsides for smoke. AI works through more than seven million generated images each day and flags anything that looks like fire to a nominated individual who then decides what happens next.

Last year, the system supported the California Department of Forestry and Fire Protection’s response to circa 3,600 wildfires. In more than half of those fire episodes, officials knew about the fire outbreaks before anyone had called 911.

Thermal imaging drones

Closer to home, United Utilities spent this summer flying thermal imaging/heat-seeking drones over moorland in North West England, subsequently finding hidden ‘hotspots’ left behind by the wildfires so that dedicated firefighting crews could then determine to put them out before they reignited. Used like that, there’s no question about it: AI is making people that much safer.

Lindsay Page, United Utilities’ lead for water services in Greater Manchester, explained: “The thermal imaging drones really are game-changing as they can pinpoint the problem areas so that we can target our resources to the right place at the right time.”

The company had already trained members of the team as drone pilots to help with monitoring assets in remote locations. The thermal heat-seeking drones were then flown to scan large areas of moorland at one time. Once the aforementioned ‘hotspots’ were found, the GPS co-ordinates could then be shared and teams sent straight out to extinguish any fires.

While projects like these inevitably grab the headlines, the most widespread use of AI in the UK’s fire safety space has arrived with no fanfare at all. It’s members of your team quietly using tools like ChatGPT without you even necessarily knowing about it.

This is what’s commonly referenced as ‘Shadow AI’: members of staff using AI tools without their employer’s approval or oversight. This isn’t peculiar to our sector. In a Microsoft survey of 2,000 UK employees conducted last year, 71% of respondents confirmed that they had used unapproved consumer AI tools at work, while more than half did exactly that every week. If you haven’t yet decided how your business uses AI, the likelihood is that someone in your team has decided for you.

Personal tools: the problem

“Unfortunately, in the fire industry there are a lot of people now using ChatGPT to write whole reports,” observed James Milton, director of TJM Fire Safety. Used in that way, a chatbot adds ‘fluff’, not substance. It wasn’t in the building, so the best it can do is turn thin notes into longer and generic text that could describe pretty much anywhere.

At worst, it risks introducing a mistake – perhaps drawn from an American fire code or otherwise out-of-date UK legislation – stated with complete confidence. It’s not just a question of quality in your work. It’s a question of data privacy. If a team member puts a company report into ChatGPT to check it over, and that report includes residents’ names or Personal Emergency Evacuation Plan details, pasting them into a free personal account is likely a breach of the UK’s General Data Protection Regulation obligations. Whatever goes wrong, it’s your consultant’s name on the report, not the software’s.

Banning AI rarely works. In the same Microsoft survey, in fact, more than a quarter of those using unapproved tools said this was because their employer offered no approved alternative.

None of this means general-purpose AI is useless. ChatGPT, Claude and the rest are all fantastic tools for lots of tasks, but in truth they’re just the tip of the iceberg. The real magic happens when AI systems are built around the specific work that’s being conducted.

Better evidence from site

Take fire risk assessments, for example. The bread and butter of so many consultancies. Much becomes possible when AI is built around the work itself.

Instead of scribbled notes or time spent filling out forms on an iPad, an AI-enabled consultant can simply talk through what they’re seeing as they walk the building, while AI sorts each observation and photograph into the right section of their report.

“Now, I just take the photographs and record my findings by speaking into an app,” noted Lee Brown (founder and director of Absolute Health Fire Safety). “What used to take us two hours per report now takes us about one.”

This isn’t just a question of producing reports more quickly. As talking is easier than writing or typing, the detail that assessors typically give using AI note taking systems goes up, not down. For an industry looking to move away from ‘box-ticking’ reports, that’s a godsend.

Much of a given fire risk assessor’s time isn’t taken up by work that requires their professional judgement. Hours every week are spent typing up handwritten notes or uploading and formatting images. Many fire risk assessors work late into the evening or at weekends catching up on paperwork, with backlogs of reports to finish from site visits they conducted perhaps several weeks ago.

When you’re seemingly drowning in paperwork, the risk is that people make mistakes or look to cut corners. When the write-up is less time-consuming, fire risk assessors have far more time to focus on the work that requires their expertise and judgement.

Second pair of eyes

AI is a tireless second pair of eyes, flagging a finding that contradicts the overall risk rating, a deficiency with no matching action or a dateline that simply doesn’t add up. Gary Mitchell (director at SMS Alba and someone who has written fire risk assessments for 15 years) noticed that AI was catching details in photos he missed the first time around. “AI noticed a certificate was out of date,” explained Mitchell. “I hadn’t spotted it.”

That doesn’t replace peer review. Instead, it actively serves to strengthen it. With the routine checks completed, a validator’s time is then spent on areas that need and demand experience: whether the risk rating is right and the recommendations are proportionate.

“You have more gateways for a check before the report’s released to the client,” affirmed fire risk assessor Duane Smith from Elite Fire Stop.

The same pattern holds across a consultancy’s wider work. Fire door inspections and compartmentation surveys generate hundreds of photos and defects that need logging, locating and turning into a schedule of works. Fire strategies have to be cross-referenced against drawings and standards. The expert work remains with the expert: the logging, formatting and checking is where AI gives the time back.

Importantly, what doesn’t change is who makes the judgement. “AI will never take over what I do,” asserted Gary Mitchell. “It doesn’t have my eyes. AI can help to analyse a photograph, but it cannot decide which photos need to be taken in the first place.”

Practical guidelines

Those businesses treading the correct path have made a decision rather than leaving it to individuals. “We want our team to be efficient, of course,” said Daniel Bright (director of CCIM, itself a nationwide consultancy), “but it’s crucial we maintain the highest standards of quality and that sensitive customer data is kept private and secure. We’ve written clear guidelines for how our team members need to use AI and given them a purpose-built platform for fire safety work. It has been transformational.”

Bright went on to state: “While we welcome the efficiencies AI brings, our approach remains firmly assessor-led and quality assurance-driven, retaining the human element and professional judgement essential for determining risk subjectively, having regard to the specific asset, its characteristics and all surrounding circumstances. AI should enhance professional expertise by working side-by-side with it.”

Any good AI policy for fire safety consultancy businesses fits on one page and should cover six principles as follows:

*AI helps with the admin, but people make the judgement. Observations, ratings and recommendations belong to a named and competent individual who reads and owns every line before anything’s sent out

*No client or resident data in personal accounts. Use approved tools only from providers who will sign a data processing agreement and will not train their models on your data

*Work from your evidence. Tools should build reports from what your people observed, not fill gaps with plausible text, and should show the origin of any reference to a standard

*Check twice before anything leaves the building. Pair automated checks with peer review so that software catches the routine errors and validators focus on the judgement calls

*Be open about it. Tell staff exactly what’s allowed and have a clear answer ready for those clients who ask the question as to how you’re using AI

*Protect the learning. New starters should learn to write a report before they learn to automate one

More work, fewer people

The case for making sure this is right goes beyond any single report. Since the Grenfell Tower tragedy, each major regulatory change has added to the sector’s workload and the bar for competence has rightly risen in parallel.

Those transacting the work are not becoming any younger. In the Home Office’s 2023 survey of fire risk assessors, 65% of them were aged 50 or older and only 8% were under 35.

Many assessors now in their sixties are choosing to retire early rather than requalify for their last few years of work. When they do retire, each one takes decades of experience with them.

Many of the sector’s most experienced professionals came to it as a second career. Lee Brown spent 27 years in the Fire and Rescue Service ahead of retraining as a consultant. Before changing how he worked, his days ran to 12 hours. “I would come home,” said Brown, “open the laptop, start uploading all of my photos and keep going until ten or eleven o’clock at night.” That simply isn’t a sustainable way in which to keep experienced people in the profession or to attract new ones into the ranks.

More work, fewer people and the requirement for a higher standard leave only one direction of travel: every firm needs to derive more from the expertise it already possesses in-house. Used well, AI is how that happens. It allows experienced individuals to take back their evenings, small businesses to take on board valuable work they would once have been forced to turn away and also frees precious time for more thorough inspections. Used carelessly, AI does the exact opposite.

Ultimately, the choice to be made by today’s company directors isn’t whether or not their team members should use AI. Rather, it’s all about the ‘How’.

Jasper Bartlett is Co-Founder of FireCheckr (www.firecheckr.co.uk)

 
OTHER ARTICLES IN THIS SECTION
FEATURED SUPPLIERS
TWITTER FEED