Introduction
Most organisations believe they understand safety because they can measure it. They track injury counts, lost-time rates, audit scores, and days without incidents. When those numbers improve, it feels like progress. Leadership attention shifts elsewhere. The assumption is simple: fewer incidents must mean lower risk. That assumption is comfortable—and incomplete. Across industries, overall injury rates have steadily declined, yet serious injuries and fatalities have not followed the same curve. This growing gap is one reason why many organisations are now exploring AI workplace safety technologies to gain earlier visibility into high-risk conditions that traditional metrics fail to capture
That assumption is comfortable. And it’s incomplete.
Across industries, overall injury rates have steadily declined. Serious injuries and fatalities have not followed the same curve. In many sectors, they have remained stubbornly flat. This gap has forced safety leaders to confront a difficult reality: reducing all incidents does not automatically reduce fatal ones.
That is the context in which the PSIF Triangle matters. pSIF stands for Potential Serious Injury or Fatality. It focuses not on what happened, but on what could realistically have happened if conditions were slightly different. And it leads to one uncomfortable truth:
A small fraction of incidents—roughly 20%—decide whether people live or die.
Why Traditional Safety Metrics Miss Fatal Risk
Traditional safety programs are built around frequency. How often do incidents occur? How many were recordable? How many caused lost time? These measures are useful for understanding general discipline and housekeeping. But they are poor at predicting catastrophic outcomes.
Minor injuries usually involve low energy and allow for recovery. Fatal incidents are different.
They almost always involve:
- High energy (height, speed, pressure, heat, electricity)
- Uncontrolled movement of vehicles or machinery
- A single failed or missing barrier
- Very little time to correct once things go wrong
This is why organisations with excellent injury statistics still experience fatalities. The system is optimised for what happens often, not for what happens when failure is irreversible.
What the PSIF Triangle Actually Shows
The PSIF Triangle is often mistaken for a variation of the classic safety pyramid. It is not. The traditional pyramid shows how frequently events occur. The PSIF Triangle shows where consequence concentrates.
At the base are:
- Unsafe acts and conditions
- Minor incidents
- Near misses
These are common, visible, and easy to count.
At the top are:
- Events with credible fatal potential
- Situations involving high energy
- Tasks where a single failure could be catastrophic
These are rare, easy to overlook, and difficult to manage with basic reporting.
The triangle narrows sharply not because risk disappears, but because fatal risk concentrates
Why Near Misses and PSIFs are Not the Same Thing
Near misses have long been treated as the gold standard of proactive safety. They matter—but they are not enough. Near misses require exposure. Someone has to be in the wrong place at the wrong time. The event must unfold. The organisation must notice it. Then it is reviewed later.
PSIFs exist before that exposure happens.
A PSIF is not defined by proximity or luck. It is defined by structure:
- High energy is present
- A direct control is missing, bypassed, or unreliable
- Failure would likely result in serious injury or death
Near misses tell us what almost happened. PSIFs tell us what must not be allowed to happen at all. In many operations today, computer vision safety alerts are being used to detect these early behavioural risks before they escalate into serious incidents.
Why PSIFs are Routinely Invisible
Most organizations already experience PSIFs. They simply don’t label them that way.
This happens for predictable reasons:
- Outcome bias: if no one was hurt, the event feels less serious
- Normalisation: risky conditions become “how work is done”
- Reporting pressure: people avoid escalation that may slow work
- Data overload: PSIFs are rare and get buried in large datasets
The absence of injury is mistaken for the absence of risk. Over time, this creates a dangerous illusion of control—until the system fails in a way that cannot be recovered. This visibility gap is one of the reasons AI workplace safety solutions are increasingly being used to detect operational risks before they escalate.
Energy, Barriers, and Why Fatalities Repeat
Serious injuries and fatalities are governed by physics, not intent.
High-energy hazards—falls, vehicles, rotating equipment, electricity, pressure, heat, chemicals—are only safe when direct controls are in place. These are controls that do not depend on perfect human behaviour.
PSIFs emerge when:
- A critical control does not exist
- A control exists but is bypassed
- A control is assumed to work but is not verified
This is why fatal events repeat around the same hazards year after year. Awareness is not the problem. Barrier reliability is.
Where Technology Becomes Essential (not optional)
PSIF signals are weak by nature. They do not announce themselves loudly.
They appear across:
- Permit deviations
- Maintenance notes
- Unsafe condition reports
- Contractor activities
- Repeated small workarounds
No individual—and no weekly meeting—can reliably connect these signals at scale. This is where specific safety technologies matter. In fact, a growing number of organisations are exploring how AI-driven computer vision is reshaping workplace safety across industries.
Modern PSIF-focused safety systems typically combine:
- Computer vision
Used to detect high-risk situations in real time by analysing live camera feeds across industrial sites. Computer vision can automatically identify workers at height without fall protection, missing PPE, unsafe pedestrian-vehicle proximity, and entry into restricted zones in real time. Platforms like Binary’s AI-enabled safety intelligence systems bring these signals together to help safety teams detect exposure early and intervene before incidents occur. It can also flag dangerous equipment interaction or personnel entering confined or high-risk operational areas without authorisation. These exposures often develop within seconds and may never appear in incident reports. By continuously monitoring existing CCTV infrastructure and generating real-time alerts, safety teams gain early visibility into high-energy risks and can intervene before a near miss—or a serious incident—occurs. - Sensor and asset data
Used to monitor equipment condition, isolation status, pressure, temperature, or movement—especially where failure would release high energy. - Digital permit and control-of-work systems
Used to ensure critical controls are verified, not assumed, before high-risk tasks begin. - Severity-weighted analytics
Used to prioritise incidents and conditions by fatal potential, not by frequency or outcome.
The value of this technology is not better reporting. It is earlier intervention—before exposure, not after luck.
What Effective PSIF Focus Looks Like in Practice
Organisations that manage PSIF risk well behave differently.
They:
- Elevate PSIFs above routine incidents, even when no injury occurs
- Review them at senior levels, not just within safety teams
- Focus investigations on barrier failure, not individual error
- Accept more “false alarms” to avoid missing a single fatal precursor
- Track the health of critical controls, not just injury trends
- Use technology to maintain visibility across sites and contractors
This is not about adding complexity. It is about directing attention to the few situations where failure cannot be tolerated.
The Leadership Question that Actually Matters
Not:
- How low is our injury rate?
- How many incidents did we log?
But: Which situations in our operations could realistically have killed someone—and what did we change because of them?
If that question cannot be answered clearly, safety is being measured, not managed.
Why 20% of Incidents Decide Life or Death
Most incidents help improve order, compliance, and discipline. A small fraction test whether the system can survive stress.
Those 20% sit at the top of the PSIF Triangle—where energy, controls, and people intersect. Ignoring them does not reduce risk. It delays consequence.
This is where platforms like Binary Safety Suite fit naturally—not as generic safety software, but as PSIF-focused safety infrastructure.
Binary helps organisations:
- Surface pSIF exposure early using computer vision and operational data
- Prioritise what matters most based on fatal potential
- Track critical control effectiveness over time
- Connect weak signals across sites and contractors
- Act before high-energy risk becomes irreversible
Not to create more dashboards—but to ensure attention reaches the risks that actually decide outcomes. Because in the end, life or death is determined by the risks you choose to see—before you’re forced to see them.
