Queensland's latest psychosocial-hazard campaign should be read as a management-system warning, not another prompt to run a wellbeing campaign.
Resources Safety and Health Queensland (RSHQ) audited 61 coal, mineral-mine, quarry, and petroleum and gas operations. Only 31 had completed a documented psychosocial hazard risk assessment. Of the operations with a completed assessment, one-third had not integrated the identified risks and controls into their safety and health management system (SHMS) or safety management system (SMS). RSHQ assessed 39 of the 61 operations as being in the early stages of developing effective capability (RSHQ campaign report).
The leadership message is direct: psychosocial risk cannot sit beside the safety system. It has to operate through it.
The gap is not awareness. It is integration.
Most audited operations were not starting from zero. RSHQ found that 87% provided an employee assistance program and 85% had proactive mental-health or wellbeing initiatives. Those supports matter, but they are not substitutes for identifying hazards, assessing risk, designing controls, consulting workers, investigating events, and reviewing whether controls remain effective.
The audit found several breaks in that chain:
- 51% of operations had completed a documented psychosocial risk assessment.
- About two in five did not use available workplace data, such as incident reports, worker surveys, and complaints, to inform the assessment.
- Of the operations with documented reporting processes, 29% had not embedded those arrangements into the SHMS or SMS.
- About half of audited operations had no documented safety and health investigation process for psychosocial reports and incidents.
- Where investigation processes existed, 81% did not include trauma-aware principles and 41% did not document how outcomes would inform ongoing risk management.
- 57% did not provide workers with training on identifying and reporting psychosocial hazards, while 49% did not provide equivalent response and management training for supervisors.
These are not isolated document defects. Together, they describe a system in which a concern may be raised but fail to become risk intelligence, a control improvement, a training change, or an accountable action.
RSHQ's regulatory response reinforces the point. Following the audits, it issued recommendations to 60 of the 61 operators. Recommendations on reporting and investigation processes went to 56 operations, and recommendations on SHMS or SMS integration went to 34 (RSHQ campaign report).
Treat psychosocial hazards like operational hazards
Psychosocial hazards differ from mobile-equipment, strata, or energy-isolation hazards, but the governance discipline should be recognisable.
A defensible system needs a documented risk assessment, controls with accountable owners, multiple trusted reporting pathways, competent and fair investigations, and a way to feed learning back into the system. It also needs consultation throughout the process, not only when hazards are first identified.
This is consistent with the broader Australian risk-management model: identify hazards, assess risk where necessary, implement the most effective reasonably practicable controls, and review whether those controls remain effective, with consultation supporting every stage (Safe Work Australia model Code). Queensland resources operations are governed by their sector-specific legislation and RSHQ guidance, so sites should confirm their precise obligations rather than assume the general WHS code applies directly.
The practical shift is from programs to control architecture.
An employee assistance program responds to people who seek support. A risk-management system asks what features of work design, supervision, rosters, accommodation, reporting access, interpersonal conduct, or organisational justice are creating exposure; which controls address those sources; and what evidence shows those controls are working.
Four tests for executive assurance
Leaders do not need to manage individual cases. They do need confidence that the system joins up.
1. Is the risk assessment connected to real evidence?
The assessment should draw on more than a workshop. Incident and complaint data, surveys, consultation, absenteeism patterns, exit themes, work design, accommodation conditions, and operational change can all provide relevant signals. Sensitive inputs require appropriate privacy and access controls, but excluding them entirely can leave the assessment detached from work as performed.
2. Can workers report through channels they trust?
RSHQ found gaps in anonymous participation and reporting. A channel may exist on paper yet remain unusable where workers fear identification, retaliation, or loss of confidentiality. Leaders should test accessibility across employees, contractors, rosters, languages, locations, and accommodation settings.
3. Do investigations strengthen risk controls?
A behavioural or HR process and a safety-and-health investigation may run alongside each other, but they do different work. The safety process must examine system conditions and feed outcomes into risk management. RSHQ specifically calls for investigations that are fair, transparent, timely, trauma-aware, and connected to ongoing risk controls (RSHQ campaign report).
4. Does change control reach every affected document and workflow?
When a control changes, the impact may extend to the SHMS, risk registers, reporting protocols, investigation procedures, supervisor training, contractor onboarding, accommodation standards, and assurance activities. Updating one policy is not system integration.
Where governed AI can help, and where it must stop
The volume and cross-document nature of this work create a legitimate role for AI. It can help reviewers compare requirements with controlled documents, identify missing links, surface contradictions, and maintain a traceable remediation list. It can also make change-impact reviews more systematic.
But psychosocial risk is a high-consequence, privacy-sensitive domain. AI must not make employment, medical, legal, disciplinary, or final safety decisions. It should not infer a person's mental state, judge credibility, or replace competent, trauma-aware investigators. MineGuard AI's published responsible-AI boundaries explicitly preserve those decisions for authorised people and require human review of AI-assisted outputs (MineGuard AI Responsible AI).
The right operating model is assistive: AI improves coverage and traceability; people retain context, judgement, consultation, and accountability.
A practical MineGuard path: start with one controlled document set
MineGuard AI's SHMS AI is designed to ingest safety management systems, procedures, risk assessments, standards, and related controlled documents; map them against nominated legislation and internal requirements; detect gaps and contradictions; and return clause-level findings with citations. Its documented workflow includes cross-document checks, version comparisons, change-impact analysis, and action-ready remediation outputs (SHMS AI).
For psychosocial-risk uplift, a focused first assessment could bring together:
- the psychosocial hazard risk assessment
- the relevant SHMS or SMS sections
- reporting and escalation procedures
- investigation procedures
- consultation arrangements
- supervisor and worker training materials
- action registers and review schedules
The aim is not to ask AI whether the operation is compliant. It is to give competent reviewers a structured, evidence-linked view of where requirements are absent, only partly addressed, inconsistent across documents, or disconnected from implementation.
MineGuard AI also offers governance and workflow consulting for high-consequence operations, including SHMS cross-document consistency, human decision boundaries, audit-trail requirements, data readiness, and governed implementation (MineGuard AI Consulting).
RSHQ's audit establishes a clear baseline. The commercially useful next step is equally clear: choose one psychosocial-risk document pack, test whether its assessment, controls, reporting, investigations, training, and review process form a connected system, and assign accountable remediation.
Book a demonstration or request a free sample SHMS AI assessment. Human reviewers remain responsible for every conclusion and action.