The failure wasn't just human — it was a protocol gap
When bodycam footage of the 2023 assault on Jodi Knott was released by ABC Four Corners on 30 May 2026, the immediate story was about two police officers — Senior Constable Nathan Black and Constable Timothy Trautsch — whose conduct a District Court judge described as "gratuitous cruelty." Knott, a woman with schizophrenia in acute psychotic distress on a Western Sydney street, died of cancer 18 months after the incident. The ABC investigation is sourced here: abc.net.au.
The public conversation is rightly focused on police training reform and the structural shortage of clinical co-responders in NSW. But if you build, run, or configure security operations — venue security, transport hubs, retail loss prevention, event staffing — there is a tighter systems question embedded in this case: at what point in your response tree does decision logic actually branch? Because in the Knott case, it didn't branch at all. There was no triage step. No "scared versus dangerous" gate. Just a default-to-control path that should have been the last resort.
That is an architectural problem, not just a training problem. And it is fixable.
The pre-arrival window is where your system's defaults matter most
Mental health crises in public spaces do not announce themselves with obvious cues. They start with someone stationary in a high-flow zone, non-responsive, or visibly dysregulated without any apparent threat vector. That window — before emergency services are on scene — is where your team's defaults do the most damage or the most good.
The worst default is crowd-in-and-contain. A person in acute psychosis or dissociative distress has a nervous system already pegged at capacity. Additional faces, movement, and noise inside their perimeter raise that load. The physiological response is not rational — it does not matter that your staff are there to help. If the sensory input says threat, the nervous system responds to threat.
The correct default is the opposite: reduce inputs. Disperse bystanders. Lower voice. Create lateral space. Do not touch unless there is an immediate physical danger. None of this is complicated. It does not require clinical credentials. It requires having made a decision in advance about what the first 90 seconds look like — before your staff are standing in the middle of a situation with no pre-loaded response pattern.
The single branch your decision tree needs
Security staff, transport workers, and venue managers encounter people in mental health distress regularly. Most have no structured response logic at all. The branch that de-escalation research — across emergency nursing, crisis counselling, and police reform literature — consistently validates is this one:
Is this person frightened, or are they actively dangerous to someone right now?
Those states can look identical from 10 meters away. Shouting, pacing, refusal to make eye contact: the body language overlaps. But the correct response to each is nearly opposite.
- Frightened: reduce pressure. Fewer bodies nearby, slower movement, lower volume, physical space, wait.
- Active danger to others: verbal de-escalation first if any window exists, then escalate response only as required.
The failure mode is skipping the branch entirely and defaulting to control in both cases. That is how welfare checks become use-of-force incidents. Building the branch into your standard operating procedure — even as a single verbal prompt before approach — changes the statistical outcome of your team's responses over time. That is not a hypothesis. It is what the co-response and de-escalation research literature shows, and it is what the Knott case illustrates in the starkest possible terms.
Practical implementation for operators running teams today
You do not need a policy rewrite cycle to get this into your operation. These are low-overhead changes:
Pre-shift brief, one minute, recurring. Run the "scared versus dangerous" question as a standing talking point for night operations and high-traffic periods. Repetition beats depth of training here. The goal is to load the decision branch into working memory before it is needed.
First responder on scene is not the decision-maker alone. One staff member creates calm space near the individual. A second calls emergency services and gives a behavioural description — what the person is doing, not an amateur diagnostic label. Additional staff stay back unless there is an active safety need. A stack of uniforms converging at once is a stimulus load, not reassurance.
Know your co-response options before the incident. In parts of NSW, police mental health intervention teams can be requested directly. If your venue or precinct has access to a co-response service, that number should live somewhere your team can reach it under pressure. Pre-loading that information is the difference between using it at the right moment and not using it at all.
XGuard is built as a real-time marketplace and dispatch system — connecting operators to deployments, and wiring decision frameworks like the one above into guard training at the platform level. The system is designed for operators who need guards to triage distress accurately before it escalates, not just respond to apparent chaos. If you are building or running security ops and want a dispatch layer that treats response logic as a first-class variable, XGuard is worth looking at.
Pro tip: Before your staff approach someone who appears to be in crisis, give them one pre-approach prompt: "Is this person scared, or are they dangerous to someone right now?" If the answer is scared, the first move is space and a lower voice, not containment. If they cannot answer the question, default to distance. Moving in fast is almost always the wrong first move — and it is nearly impossible to walk back once the person's nervous system has registered it as a threat.
What this case asks of people who design security systems
The structural failures in the Knott case — inadequate co-response coverage, no clinical backup, officers responding alone to a medical situation — are real and documented. Her family has called for structural change, not just individual accountability. NSW needs faster investment in after-hours mental health capacity and clearer dispatch protocols.
That work is slow. Crises are not.
The operators, venue managers, and security builders closest to these moments are not going to have a clinician in the room every time. What they can have is a response tree with the right branch in it, loaded into their team before it is needed. That does not replace a co-responder. It changes what the situation looks like when one arrives.
If you are building in this space, that is the lever worth pulling first.
Check out XGuard if you are an operator looking for a dispatch and marketplace platform that treats response logic as infrastructure, not an afterthought.
Source: ABC News Australia — 2026-05-30
Originally published at marketplace.xguard.app. This version was adapted for this platform's audience; the canonical original lives at the link above.
Top comments (0)