Safety Responsibility Explained: 3 Accountability Levels

A manager can be responsible for safety and still be unable to stop a job, approve a deviation, or secure the control that prevents a serious injury. That gap is not a wording problem. It is a leadership design problem.

Safety responsibility is useful only when the organization can show who has authority, who owns the control, and who must prove that the decision was carried through. This distinction matters most during abnormal work, contractor activity, and production pressure.

Safety responsibility is the operating relationship between a decision maker, a risk owner, and the evidence required to show that agreed controls remain effective. It is not a single duty carried by the EHS function. It is a chain of authority and follow-through that must remain visible before, during, and after high-consequence work.

Definition

Responsibility describes the duty to act, while accountability describes the obligation to explain the decision and its result. Authority is the permission to make or stop that decision. When these 3 ideas are merged into one job description, the organization can assign blame without assigning control.

ISO 45001, published in 2018, expects leadership involvement and defined responsibilities, yet a compliant chart does not prove that a supervisor can challenge a permit or that a control owner can obtain resources. James Reason's work on latent failures points to the same practical issue, because the conditions around a decision can make a reasonable person appear to be the only cause of failure.

3 accountability levels

1. Authority decides what may proceed

Authority is the right to set a boundary, approve a change, or stop work when the risk exceeds the agreed limit. It normally sits with the person who controls the operating decision, not automatically with the person who advises on safety.

A shift manager may hold authority for a restart, while an EHS adviser has authority to challenge the evidence and escalate the concern. If the adviser can identify a gap but cannot trigger a review, the organization has consultation without decision power. The first practical test is simple: ask which role can say 'not yet' without seeking permission from the person under schedule pressure.

For a deeper treatment of high-consequence decision rights, see the explanation of safety decision rights.

2. Ownership keeps the control alive

Ownership is the named duty to maintain a barrier or operating condition after approval has been granted. The owner does not merely sign a risk assessment. The owner confirms that the control has a standard, a competent person, a verification rhythm, and a response when the control fails.

This is where many programs become administrative. A permit may identify an isolation, but nobody owns the test that proves the isolation remains intact after a shift change. A procedure may require a rescue plan, but nobody owns the equipment check that shows the plan is usable today. The control owner therefore needs 4 visible elements: the boundary, the evidence, the review cadence, and the escalation route.

Andreza Araujo's work on safety culture emphasizes that stated values do not become operating culture until they change decisions in the field. The same principle appears in Safety Culture: From Theory to Practice, where leadership is treated as a pattern of choices rather than a message on a wall.

3. Accountability proves what happened

Accountability begins after the decision, but it should be designed before the work starts. The accountable person must be able to answer 5 questions: What was known, what was accepted, which control was relied upon, what changed, and what was done when the change appeared'

This level is not a search for a convenient individual to blame. It is a test of whether the organization can reconstruct the decision with enough precision to improve the next one. A missing signature may be a record defect, while a missing decision trail can reveal that nobody had the mandate to challenge the work.

Use the same logic during a leadership visit to high-risk work. The useful question is not whether the leader was visible. It is whether the visit changed an owner, a control, a resource, or a decision boundary.

How to differentiate the 3 levels in practice

Test one non-routine task with a supervisor, a control owner, and the manager who approves the work. Write each role beside one decision, one control, and one piece of evidence. If the same person appears in all 3 places, confirm that the person has the time, competence, and authority to carry the whole chain.

A newly appointed supervisor can pressure-test this structure before taking a shift by using a procedure review that exposes unclear ownership. The review should take less than 30 minutes, because a responsibility model that requires a meeting to understand will fail during an upset condition.

When the model prevents leadership theater

The model is most valuable when production is late, a contractor is waiting, or a temporary control has lasted longer than planned. Those moments expose whether safety responsibility is real. A leader who owns the result but cannot change the resource plan is accountable without authority. An EHS professional who can stop a job but cannot secure a permanent fix has authority without ownership. Both conditions create frustration and weak follow-through.

Across 25+ years of executive EHS work, Andreza Araujo has built her public position around the difference between declared commitment and operating behavior. The practical lesson is direct: do not ask whether safety is everyone's responsibility until the organization can name whose decision, whose control, and whose evidence will be reviewed.

Start with one high-risk task this week. Map the 3 levels, record the 4 proof points that matter most, and remove any responsibility that exists only in a policy sentence.