Australia has required large companies to report on modern slavery in their supply chains since 2019 with no penalty attached to not reporting — a statutory review recommended civil penalties in 2023 and none was law on 2026-08-17

What is happening?

The Modern Slavery Act 2018 of the Commonwealth commenced on 2019-01-01. It requires every entity carrying on business in Australia with consolidated revenue of at least 100 million Australian dollars to publish an annual statement describing the risk of modern slavery in its operations and supply chains, and what it has done about that risk. The reporting obligation itself carries no civil and no criminal penalty. Because the Act provides none, there is no penalty available to impose on an entity that does not lodge a statement, or that lodges one that is inaccurate.

The regime does produce a public record. The government register held 16,446 mandatory statements and 1,097 voluntary statements at 2026-08-17, and records 28,108 entities named as covered across all statements lodged since the first reporting cycle in 2019. That last figure is cumulative over seven years rather than a count of entities reporting in any one year.

The Act carried its own three-year check. A statutory review commenced 2022-03-31 and its report was tabled in Parliament on 2023-05-25, prepared by an external reviewer with support from the Attorney-General's Department. That review reported on 2023-05-25 that the reporting-only approach of the Act had not always translated into accountability, and it made 30 recommendations. Three of them go to the gap described here: civil penalties for failing to report, for a materially false statement, and for not complying with a remediation request · lowering the reporting threshold from 100 million to 50 million Australian dollars, which would have brought a further 2,393 entities into the regime · and creating an Anti-Slavery Commissioner.

The government responded on 2024-12-02 and accepted 25 of the 30 recommendations in full, in part or in principle. On the same day it created the office of Anti-Slavery Commissioner and appointed the first holder of that office, without investigation or enforcement powers. It did not adopt the civil penalty recommendation, deferring it to further consultation, and it expressly rejected the reduction of the threshold to 50 million Australian dollars.

On 2026-07-16 the federal government announced an intention to amend the Act — a new criminal offence of failing to take reasonable steps to address modern slavery risk in supply chains, with a defence of having taken reasonable steps · civil penalties for breaches of the existing reporting obligation · and powers for a regulator to gather information, accept enforceable undertakings, issue infringement notices and apply for civil penalties. At the time of that announcement the amendments had not been introduced as legislation, and the government stated that the detail remained subject to further consultation. Seven years and seven months after the obligation commenced, the consequence attached to it is still the same one it started with, which is none.

Whose problem is this?

The body that would have to attach a consequence is not the body that administers the Act, and the office created to watch the regime was given no power to act on what it sees. Nothing inside the system can convert an unmet obligation into a cost.

Where does this problem end?

Three value questions sit immediately next to this document and none of them is decided in it. Whether mandatory human rights due diligence should be legislated in place of disclosure is one. Whether import controls on goods made with forced labour would do more than any reporting regime is the second. Whether the threshold should sit at 100 million or 50 million Australian dollars is the third, and on that one the review and the government reached opposite conclusions, both of which are recorded above. This document measures the distance between an obligation and a consequence, and takes no position on which instrument ought to close it.

What is the state now, and what should it be?

Now

The record therefore contains a great deal of visible activity. A review was commissioned on time and reported on time, most of its recommendations were accepted, an office was created and filled, a register publishes thousands of documents, and an amendment has been announced. The one thing none of that has produced is a consequence for an entity that does not report.

Needs a new measurementthe target state: no source opened here names a date by which penalties are meant to exist, a compliance rate the regime is meant to reach, or a standard of statement quality the Act is meant to produce. The statutory review named the instruments it thought were missing and the announcement of 2026-07-16 named the instruments the government intends to create, and neither carries a commencement date or a number that a later year could be measured against. Nobody is failing to hit a figure here, because no figure has been set.

How big is it?

The population layer for this document records not-derivable, and the reason is that no source opened here converts a disclosure gap into a count of people. The quantities that exist measure three different things and none of them is a headcount of the affected.

The regime. The register named 28,108 entities as covered across all statements lodged since the first reporting cycle in 2019, and held 16,446 mandatory and 1,097 voluntary statements at 2026-08-17. Those are counts of entities and of documents, accumulated over seven reporting years, and they are not a count of entities obliged to report in a single year. No source opened here gives that second number.

The exposure the regime is aimed at. An analysis published on 2026-02-03 by two non-government research organisations estimated that Australia imported close to 100 billion Australian dollars of goods at high risk of forced labour during 2024, which those organisations put at about one fifth of all imports that year. That is a value of trade rather than a number of workers, and it stands as a proxy for the size of the economic surface the reporting obligation is meant to make visible.

The domestic prevalence figure. An international anti-slavery research foundation estimated 41,000 people in modern slavery in Australia on any given day in 2021, an increase of 173 percent on its own previous estimate for 2018. That figure is a national prevalence estimate. It is not confined to the supply chains the reporting obligation covers, it was produced by a non-government body rather than by a statistical agency, and the index page carrying it did not open in this round, so it reaches this document at second hand.

Statement quality is known only from a sample. A 2023 study by a civil society research collaboration reviewed 102 statements drawn from four high-risk sectors — garments sourced from China, rubber gloves from Malaysia, seafood from Thailand and fresh produce grown in Australia — and reported that 77 percent did not meet the basic reporting requirements, that 52 percent did not identify obvious modern slavery risk, and that 73 percent showed no evidence of effective action. Those statements were selected by sector rather than at random and were chosen because the sectors carry high risk, so the percentages describe that sample and cannot be extended to the whole register. No later source opened here repeats the exercise at any larger scale.

Under what conditions does it arise?

1. The obligation and the consequence were separated when the Act was made. The duty to lodge a statement exists, the register publishes what is lodged, and nothing in the statute converts a missing or hollow statement into a cost. An entity that reports fully and an entity that does not report at all face the same legal position, which is that nothing happens.

2. The record measures lodgement rather than content. The register counts statements and the entities they name. No source opened here describes the administering body assessing whether a lodged statement meets the mandatory criteria, and the only quality figures available come from a sample assembled outside government. A regime whose only published measure is how many documents arrived cannot distinguish between compliance and paperwork.

3. The office created to watch the regime cannot act on what it sees. The Anti-Slavery Commissioner was established on 2024-12-02 and given no investigation and no enforcement power. The estimate that between a few hundred and more than a thousand covered entities do not report comes from that office, which means the regime can describe its own shortfall and can do nothing about it.

4. Nobody has published who is supposed to report. The threshold is a revenue test, so the population of covered entities is determined by the accounts of each entity rather than by a list held by a regulator. No source opened here states that the government maintains or publishes such a list. Where the denominator is unknown, a non-reporting rate cannot be computed, and an obligation whose breach cannot be counted cannot be shown to be breached.

5. A review with recommendations is not a deadline. The Act required the review, the review reported on time, the government answered it, and the answer deferred the penalty question to further consultation with no date attached. The announcement of 2026-07-16 did the same thing at one further remove, naming instruments without naming when they begin. Each step is a real step and none of them is a commitment that a later year can be measured against.

What has been tried?

Two directions have been tried in sequence rather than at once — build the disclosure record first, and consider consequences afterwards. The first has run for seven years and produced a large public archive. The second has been recommended, deferred, and announced, and at 2026-08-17 it had not produced a provision that an entity could breach.

What was found?

Why is it still unsolved?

Enforcement absent — the duty exists, the record of who has answered it is public, and the statute provides nothing that follows from not answering it.

A duty that carries no consequence is a request that has been written down. That is the whole of the first reason, and it is a design choice made in 2018 rather than a failure of administration since. The department that runs the register can publish what arrives. It cannot make an entity lodge, it cannot reject a statement as inadequate, and it has no instrument to reach an entity that stays silent. Seven reporting cycles have passed under that arrangement and the position at 2026-08-17 is identical to the position at 2019-01-01.

The second reason is that the regime measures the wrong quantity for its own purpose. The register counts statements lodged and it does not count statements that are true. A count of documents rises whenever more documents arrive, and it would rise in exactly the same way whether the underlying due diligence deepened or stopped. There is no published assessment by the administering body of how many lodged statements meet the mandatory criteria, so the only visible number is one that cannot fall when quality falls.

The third reason is that the shortfall cannot be sized from inside government. The threshold is a revenue test, which means the set of entities obliged to report is defined by their own accounts rather than by a register of the obliged. No opened source states that any list of covered entities is maintained or published. The one figure that exists for non-reporting is a range wide enough to differ by an order of magnitude, and it was offered by an office that has no power to check it. A breach that nobody counts is a breach that nobody has to answer for, and that holds independently of whether a penalty is ever created.

The fourth reason, which is the one that keeps the pattern stable rather than merely slow, is that every step taken so far is a real step that changes nothing about the consequence. A statutory review was required and delivered. Most of its recommendations were accepted. An office was created and filled. An amendment has been announced. Each of those is genuine activity and each is reported as progress, and none of them attaches a cost to not reporting. A gap that keeps producing visible motion produces no pressure, because from outside it looks like something is already happening.

What observation would mean it is solved?

Candidates — (a) penalty provisions commence, with a stated effective date and a body empowered to apply them (b) the government publishes a count of entities obliged to report and a count of those that lodged, and the second rises toward the first (c) the administering body assesses lodged statements against the mandatory criteria and publishes the result, and the share meeting them rises.

(a) alone is weaker than it looks. A penalty can arrive with a threshold high enough, a defence broad enough or a commencement distant enough that no entity is ever within reach of it. The announcement of 2026-07-16 already contains a defence of having taken reasonable steps, and what counts as reasonable is exactly the question the regime has not been able to answer for seven years. A provision that exists and is never applied would satisfy this observation while leaving the position unchanged.

(b) is the observation this document would most want and the one nothing currently supports. Without a denominator there is no rate, and no opened source shows that anyone holds the denominator. Building it would require identifying every entity above a revenue threshold, which is a different administrative task from running a register of what arrives. It is also the observation most easily faked in the wrong direction, because a published denominator that is too small makes compliance look complete.

(c) alone counts documents against a checklist. Meeting the mandatory criteria is not the same as finding forced labour in a supply chain and doing something about it, and the sampled study that exists already separates those two things by reporting that 52 percent failed to identify obvious risk while a larger share failed the formal requirements. A rising score against criteria could reflect better drafting rather than better due diligence. The three have to be read together, and (b) is the one that would show whether the obligation binds anybody at all.

What is it connected to?

Fills with researchstate and territory modern slavery legislation and the separate commissioner and procurement regime it creates, import restrictions on goods produced with forced labour, mandatory human rights due diligence regimes in other jurisdictions and how their enforcement provisions compare, the trafficking and forced labour offences in Divisions 270 and 271 of the Commonwealth Criminal Code, and corporate disclosure obligations in Australia that do carry penalties. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• How many entities are obliged to report. No source opened here gives a government count of entities above the reporting threshold. The 28,108 figure on the register is a cumulative count of entities named as covered by statements since 2019, which is a different quantity, and no opened source converts one into the other.
• How many entities fail to report. No source opened here carries a figure published by the government. The range of a few hundred to more than a thousand is attributed to the Anti-Slavery Commissioner in a legal practice note of 2025-11-06, and that note does not state how the range was produced or the date it describes.
• Why the threshold recommendation was rejected. The government response of 2024-12-02 records the rejection of the reduction to 50 million Australian dollars. No source opened here carries a reason given for it.
• When the announced penalties would begin. The announcement of 2026-07-16 is described as an intention with detail subject to further consultation. No opened source names a bill, an introduction date or a commencement date, and none reports any assessment of how many entities the announced offence would reach.
• Whether the sampled compliance figures hold across the register. The figures of 77, 52 and 73 percent come from 102 statements chosen in four high-risk sectors. No opened source repeats the exercise across the full body of lodged statements, and none states what a representative sample of the register would look like.
• What the administering body has published about statement quality. No source opened here carries a government assessment of how many lodged statements meet the mandatory criteria, or any statement of what the department does when a statement is inadequate.
• What the register annual report contains. The annual report on implementation of the Act is published as a PDF on the register and did not open in this round, so whatever compliance figures it may carry are absent from this document.
• The primary documents did not open. The review report page and the Commissioner publications page returned timeouts, one legal summary returned a refusal, and one report is a binary PDF that did not parse. Five rows below are blank with the reason recorded rather than filled with a substitute link. Where an official document could differ in scope or wording from the secondary account of it, this document has no way to detect the difference, and the findings graded medium above are the ones that depend on that.

See the evidence

No primary government document was read in full. The register opened directly and supplies the statement and entity totals, and the departmental consultation page opened directly and supplies the review dates. Everything else about what the review said, what the government answered and what was announced in July 2026 reaches this document through professional summaries that cite those documents. The review report itself, the commissioner position paper, the human rights commission commentary and the register annual report did not open, so five rows above are blank with the reason recorded rather than filled with a substitute link. Where sources overlap they agree: the absence of a penalty appears in three opened sources, the threshold of 100 million Australian dollars appears in three, and the date of the government response appears in two. Where a figure rests on one source only it is graded medium or low above rather than presented as settled, and the two figures produced outside government — the prevalence estimate of 41,000 and the import estimate of close to 100 billion Australian dollars — are attributed to the kind of body that produced them rather than treated as official statistics. The compliance percentages come from a sample of 102 statements chosen in high-risk sectors and are not extended beyond that sample anywhere in this document. This is a Path A output, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below