AI for Modern Slavery Due Diligence
AI, Modern Slavery and the Australian Supply Chain
For the first six years of its life, the Modern Slavery Act 2018 was a disclosure regime with no teeth. Reporting entities lodged a statement each year, the statement went on a public register, and nothing much happened if it was thin, late or copied from last year. That era is ending. Australia now has its first Anti-Slavery Commissioner, the Government has agreed in principle to introduce penalties, and large buyers are pushing due diligence expectations down onto their suppliers whether or not those suppliers are legally required to report.
For midsize Australian businesses (roughly 50 to 500 employees), this creates a two-sided pressure. If your consolidated revenue crosses the reporting threshold, your statement now has to withstand scrutiny it never faced before. And even if you sit below the threshold, you are increasingly being asked to answer modern slavery questionnaires as a condition of supplying larger customers.
Mapping a real supply chain, across dozens or hundreds of suppliers and their suppliers, is a data problem before it is a compliance problem. That is where AI can help. It is also an area where AI can produce confident, wrong answers about human beings, so the limits matter as much as the capability. This guide is for the procurement lead, sustainability manager, General Counsel or CFO who has to make that call.
What has changed in the regime
- Australia appointed its first Anti-Slavery Commissioner, whose five-year term commenced on 2 December 2024 under the Modern Slavery Amendment (Australian Anti-Slavery Commissioner) Act 2024.
- Following the 2024 statutory review, the Government has agreed in principle to civil penalties for failing to report, providing false information, or ignoring remediation requests, and consulted on the design through 2025.
- The mandatory reporting threshold stays at $100 million consolidated revenue; the Government noted, but did not adopt, a recommendation to lower it to $50 million.
- The Commissioner is expected to be able to declare high-risk regions, products, suppliers or supply chains, sharpening where scrutiny falls.
From tick-box to due diligence
The direction of travel is a move from disclosure to genuine due diligence. Under the original regime, an entity described its risks and what it was doing about them. The reforms and the Commissioner's remit push toward evidence that you actually looked, actually found risk where it exists, and actually did something about it.
Modern Slavery Compliance: The Shift Underway
| Metric | Original regime | Direction of reform | Improvement |
|---|---|---|---|
| Consequence for a weak statement | Reputational only | Penalties agreed in principle | Enforceable |
| Oversight | Register, limited follow-up | Dedicated Anti-Slavery Commissioner | Active |
| Expectation | Describe your risks | Evidence real due diligence | Higher bar |
| High-risk focus | Self-defined | Commissioner may declare high-risk areas | Directed |
| Who is asked | $100m+ entities | Their suppliers too, by contract | Wider reach |
That last row is the one that catches midsize firms by surprise. You may never lodge a statement yourself, but if you supply a bank, a retailer, a miner or a government agency, their modern slavery obligations become your questionnaire to complete. The threshold defines who reports. It does not define who gets asked.
Why supply chain mapping is a data problem
The reason modern slavery compliance is hard is that the risk usually does not sit with your direct suppliers. It sits two, three or four tiers down, in the cotton, the cobalt, the seafood, the electronics components or the labour hire that feeds your direct suppliers. A tier-one supplier in Melbourne can look immaculate while sourcing from a high-risk region you have never heard of.
Building that picture means pulling together supplier lists, spend data, country-of-origin information, product categories and known risk indicators, and keeping it current as suppliers change. Done manually in spreadsheets, it is slow, stale within a quarter, and impossible to audit. This is exactly the kind of fragmented, multi-source data problem where AI earns its place, the same pattern we describe in supply chain visibility and AI disruption prediction.
AI-Assisted Modern Slavery Risk Mapping
The value here is not that AI knows which supplier uses forced labour. It cannot know that. The value is that AI can turn a messy pile of procurement data into a ranked, current, auditable risk map, so your limited human effort goes to the suppliers that actually warrant it, rather than being spread evenly and thinly across all of them.
Where AI genuinely helps
1. Consolidating fragmented supplier data. Most midsize businesses do not have a clean supplier master. Spend sits in the finance system, contracts sit in a drive, onboarding forms sit in email. AI can reconcile these into a single supplier view, matching entities across inconsistent names and reference codes. This is the same consolidation challenge covered in AI vendor onboarding and supplier automation.
2. Risk-scoring at the category and geography level. By combining a supplier's sector, country of operation and product categories with published, credible risk indicators, AI can produce an inherent-risk score that tells you where to look first. The Global Slavery Index and government guidance provide the risk signals; AI applies them consistently across your whole supplier base instead of relying on whoever happens to know a given region.
3. Questionnaire triage and analysis. Sending every supplier a 60-question modern slavery survey annoys everyone and buries the real signal. AI can tailor questionnaires to the risk tier, then read the responses at scale, flagging inconsistent or evasive answers for human follow-up rather than leaving them in an unread inbox.
4. Statement drafting support. For entities that do report, AI can assemble first drafts of the factual sections of a modern slavery statement, structure, operations, supply chains, actions taken, from source documents. Your compliance owner then verifies and completes it. As with any regulated document, the human signs, not the model.
Where AI Shifts the Effort
The through-line is the same as everywhere else AI touches compliance. It compresses the data work so your people can spend their time on the judgement work: deciding what a red flag means, engaging a supplier, and choosing what to do about a genuine risk.
Where AI creates real risk
Modern slavery is a subject where a wrong AI answer has consequences for real people, and a false clean bill of health is worse than no assessment at all. The limits are not optional caveats. They are the reason a human stays accountable.
Inherent risk is not actual harm. An AI risk score tells you a supplier operates in a high-risk sector or geography. It does not tell you that supplier is exploiting anyone, and a low score does not mean a supplier is clean. Treating the score as a verdict, rather than a prioritisation tool, produces exactly the false comfort the regime is trying to eliminate.
Garbage in, confident garbage out. If your supplier data is incomplete, and most is, the AI will still produce a tidy-looking risk map with the gaps invisible. The suppliers you never captured do not appear as risks; they appear as nothing. Someone has to own the completeness of the input, or the output is a well-formatted illusion.
Privacy and worker data. Modern slavery due diligence can involve personal information about workers, including vulnerable people. Processing that through AI is handling personal information under the Privacy Act 1988, and worker data in particular demands care. The controls we set out in the Privacy Act compliance guide for AI apply directly.
Data sovereignty in your supplier data. Supplier lists, pricing and risk assessments are commercially sensitive. Feeding them into consumer AI services, where they may be retained or used for training, is the failure mode we describe in how the wrong AI tools leak business data. Modern slavery data belongs in a private, access-controlled workspace, full stop.
A statement is a representation. A modern slavery statement is a public document that makes claims about what your business did. Overstating your due diligence, especially with AI-generated language that sounds thorough but is not backed by real work, risks misleading conduct exposure under the same Australian Consumer Law principles that govern any other public claim. Our ACCC consumer law and AI analysis explains why the words in a statement carry legal weight.
A defensible rollout
You do not need a large team or a big budget to put a credible AI-assisted process in place. You need a sequence that produces evidence at each step, so that if the Commissioner, a customer or a court ever asks, you can show your working.
Building AI-Assisted Due Diligence
The order matters. Businesses that jump to statement-writing before the data and risk-scoring work is done end up describing a due diligence process they have not actually run. The reforms are specifically designed to expose that gap.
Do you even need to do this? A triage prompt
The threshold question is different from the merger-control style tests, because being below the reporting threshold does not mean you are off the hook. Use the prompts below to work out how hard you need to look, then confirm with your advisers.
How Hard Do You Need to Look?
Most midsize businesses fall into the middle two rows: not legally required to lodge a statement, but increasingly required by their customers to demonstrate that they have looked. That is a commercial reason to build the capability, quite apart from the law.
Where this sits in your ESG and governance picture
Modern slavery does not live alone. It sits alongside your climate reporting, your broader ESG obligations and your general AI governance. The same supply chain data that feeds a modern slavery risk map also feeds Scope 3 emissions work under the climate disclosure regime, which we cover in AASB S2 climate reporting for Group 2 entities. Building the supplier data foundation once, and using it across both, is how midsize firms avoid running three overlapping compliance projects in parallel.
The discipline of automating ESG and supply chain data credibly, rather than generating plausible-looking output, is one we have direct product experience with through Carbonly, the emissions and compliance automation platform built by our founder. That work, documented in our Carbonly case study, reinforced a lesson that applies squarely to modern slavery: the hard part is never the AI, it is the trustworthiness of the underlying data and the human accountability wrapped around the output. And all of it should sit inside a defined AI governance framework, so that who approves, who checks and who signs is written down before the first supplier is scored.
What to do this quarter
- Consolidate your supplier data first. No risk map is better than its inputs. Get suppliers, spend and categories into one auditable view before anything else.
- Use a private AI workspace. Supplier and worker data never goes into consumer AI tools. Confirm retention and training settings contractually.
- Score to prioritise, not to conclude. Treat AI risk scores as a way to direct human attention, never as a verdict on whether exploitation exists.
- Keep a human accountable at every step. Someone named owns data completeness, someone named owns remediation decisions, and someone named signs any statement.
- Build once, use across ESG. The supplier foundation you build for modern slavery also serves climate and broader supply chain work.
The businesses that come out of this reform period well are the ones that treat AI as a way to actually see their supply chain, not as a way to generate paperwork that describes a due diligence process they never ran. The regime is getting better at telling the difference, and so are the customers doing the asking.
Related Reading:
- Supply chain visibility and AI disruption prediction - The multi-source data foundation modern slavery mapping shares.
- AI vendor onboarding and supplier automation - Building the clean supplier master this depends on.
- AASB S2 climate reporting for Group 2 entities - The Scope 3 work that shares your supplier data.
- ACCC consumer law and AI for Australian midsize business - Why the claims in a public statement carry legal weight.
- AI governance framework for Australian midsize business - The structure supplier-risk AI sits inside.
Sources: Modern Slavery Act 2018 (Cth); Modern Slavery Amendment (Australian Anti-Slavery Commissioner) Act 2024 and Attorney-General's Department material on the Anti-Slavery Commissioner (term commenced 2 December 2024); 2024 statutory review of the Modern Slavery Act and the Government's response agreeing in principle to civil penalties and retaining the $100 million reporting threshold; Attorney-General's Department consultation on penalties and due diligence (2025); Walk Free Global Slavery Index risk indicators; Privacy Act 1988 and OAIC guidance on personal information and AI. Solve8 synthesis informed by ESG and compliance automation product experience. This article is general information, not legal advice.