AI for Australia's New Merger Regime
AI and Australia's New Mandatory Merger Regime
On 1 January 2026, the biggest change to Australian competition law in a generation took effect. The old informal clearance process, where an acquirer could quietly complete a deal and deal with the Australian Competition and Consumer Commission later if it took an interest, is gone. In its place sits a mandatory, suspensory notification regime run by the ACCC. If your acquisition crosses the thresholds, you must notify, and you cannot complete until you are cleared.
For midsize Australian acquirers (roughly 50 to 500 employees, or private equity and family offices operating at that scale), this is not an abstract policy debate. It changes deal timelines, it changes the cost of getting due diligence wrong, and it introduces a new category of question that has to be answered before signing: does this transaction, on its own or stacked on top of what we have already bought, trigger a mandatory filing?
AI can genuinely help with parts of that problem. It can also create fresh legal exposure if you point it at a data room without thinking. This guide is for the CFO, corporate development lead, General Counsel or deal principal who wants to use AI to move faster through diligence without tripping over the new rules or the ACL.
What actually changed on 1 January 2026
- The regime is mandatory: notification is compulsory when monetary thresholds are met, not a strategic choice.
- The regime is suspensory: you cannot complete a notifiable acquisition until the ACCC clears it or grants an exemption.
- Businesses could notify voluntarily from 1 July 2025 during the transition; mandatory notification applies to acquisitions from 1 January 2026.
- Further asset acquisition thresholds and voting power triggers commence from 1 April 2026, widening what counts as notifiable.
The old world versus the new world
The shift is easiest to understand as a before and after. Under the previous arrangements, most deals never went near the ACCC, and the regulator had to prove a substantial lessening of competition if it wanted to intervene. Under the new arrangements, notification is compulsory above the thresholds, completion is frozen until clearance, and the burden of the analysis has moved onto the parties and their advisers.
Merger Control: Before and After 1 January 2026
| Metric | Old informal regime | New mandatory regime | Improvement |
|---|---|---|---|
| Notification | Voluntary in practice | Mandatory above thresholds | Compulsory |
| Completion | Could complete, then defend | Suspended until ACCC clears | Frozen |
| Who carries the analysis | ACCC had to act | Parties must self-assess and file | Shifted |
| Serial acquisitions | Assessed deal by deal | Aggregated over three years | Cumulative |
| Cost of getting it wrong | Unwinding risk | Penalties up to $50m for completing early | Higher |
The penalty point matters. Completing a notifiable acquisition without clearance (often called gun-jumping) is a breach of the Competition and Consumer Act 2010, and the penalties reported for corporations reach up to $50 million, with lower but still serious figures for individuals. That is a board-level risk, not a footnote for the legal team.
Where the thresholds actually bite
The reason AI matters here is that the thresholds are cumulative and fact-heavy, which is exactly the kind of problem that gets missed when a corporate development team is moving quickly. Notification is required, broadly, where a transaction has a connection to Australia and any of the following is met.
- Combined size test. The combined Australian turnover of the parties is at least $200 million, and either the target's Australian turnover is at least $50 million or the global transaction value is at least $250 million.
- Very large acquirer test. The acquirer group's Australian turnover is at least $500 million and the target's Australian turnover is at least $10 million.
- Serial or creeping acquisitions. Where the combined Australian turnover is at least $200 million, notification is required if at least $50 million of relevant acquisitions have been made over the past three years. Where the acquirer's Australian turnover is at least $500 million, that three-year cumulative trigger drops to $10 million.
Read that last point again, because it is where midsize acquirers get caught. A roll-up strategy, buying five or six small competitors or suppliers over three years, can cross a notification threshold even though no single deal looks large. The regime is explicitly designed to catch that pattern. If your growth model is acquisition-led, you now need a running tally of what you have bought and where the overlaps sit.
These are the thresholds as reported by competition law practitioners at commencement. Treasury has also signalled adjustments through 2026, including the new asset and voting power triggers from 1 April 2026, so the numbers above are a starting point for a conversation with your competition counsel, not a substitute for it.
Where AI genuinely helps
None of this makes AI a decision-maker. The ACCC will not accept "our model said it was fine," and no General Counsel should sign a filing on the strength of an unreviewed AI summary. But there are four parts of the deal process where AI reduces genuine drudgery and, more importantly, reduces the chance that a human misses something in a 4,000-document data room at 11pm.
AI-Assisted Diligence Under the New Regime
1. Contract and data room review at scale. A commercial data room for a midsize acquisition routinely runs to thousands of contracts, leases and permits. AI document review can extract change-of-control provisions, assignment restrictions, exclusivity clauses, most-favoured-nation terms and material adverse change definitions far faster than a junior lawyer reading sequentially. This is the same capability we cover in AI contract review for midsize legal and procurement teams, applied to the pressure of a deal timetable.
2. Overlap mapping for competition analysis. The substantive question the ACCC asks is whether the deal substantially lessens competition. That analysis starts with mapping where the acquirer and target actually compete, by product and by geography. AI can accelerate the first-pass mapping across product catalogues, customer lists and site locations, giving competition counsel a structured starting point instead of a blank page.
3. Threshold and creeping-acquisition triage. This is the newest and most valuable use. An AI workflow that holds a structured record of your last three years of acquisitions, their Australian turnover contributions and their overlap categories can flag when a proposed deal is approaching a cumulative trigger. It does not make the legal call. It stops the deal team from discovering the problem after signing.
4. Notification drafting support. The ACCC notification form requires structured information about the parties, the markets, the rationale and the competitive effects. AI can assemble first drafts of the factual sections from source documents, which counsel then verifies and completes. The lawyer still owns the filing. The blank-page time disappears.
Where AI Saves Time in a Mid-Market Deal
The point of that last row is that AI shifts where your expensive human hours go. It does not remove them. It moves your lawyers from reading every page to verifying flagged pages, and it moves your deal team from reconstructing acquisition history under time pressure to consulting a record that was already maintained.
Where AI creates fresh risk
This is the part that gets skipped in vendor pitches, and it is the part that protects you. Pointing a general-purpose AI tool at deal data is one of the faster ways to create a legal problem.
Confidentiality and clean teams. Data rooms are governed by confidentiality agreements and, in competitively sensitive deals, by clean team protocols that restrict who can see what. Feeding that material into a consumer AI service, where the data may be retained or used to train a model, can breach those obligations outright. Any AI used in diligence has to run in a private, access-controlled environment with contractual guarantees on data handling. We wrote about the underlying failure mode in how the wrong AI tools leak business data into training.
Privacy Act exposure. Data rooms are full of personal information: employee records, customer databases, contact lists. Processing that through AI is handling personal information under the Privacy Act 1988, with all the security, purpose and disclosure obligations that follow. This does not stop you using AI. It means the workspace, the vendor and the retention settings all have to satisfy the Australian Privacy Principles. Our Privacy Act compliance guide for AI sets out the controls.
Legal professional privilege. Diligence findings are often privileged. Routing them through a third-party AI service can, depending on how the service is configured, put privilege at risk. This is a question for your General Counsel before the tool is switched on, not after.
Hallucinated clause interpretation. AI document review is fast, not infallible. It can misclassify a clause, miss a carve-out, or confidently summarise a provision that says the opposite of what it claims. In a competition filing or a purchase agreement, that is not a rounding error. Every AI flag has to be verified against the source document by a human who is accountable for the conclusion. The ACCC's own consumer law expectations are a reminder that "the AI said so" is not a defence anywhere in Australian law.
A realistic deal timeline under a suspensory regime
The suspensory nature of the new regime is what most changes deal planning. You cannot complete until you are cleared, so clearance time is now on the critical path. A well-run process front-loads the diligence and the threshold analysis so the filing is ready early, rather than treating competition clearance as a formality at the end.
Mid-Market Deal Under the 2026 Regime
The threshold assessment moving to the front is the structural change. Under the old regime you could often leave competition questions late. Under the new one, a deal that turns out to be notifiable after you have committed to a timetable can blow that timetable apart. This is precisely where a maintained AI record of your acquisition history earns its keep.
Should you even be worried? A triage prompt
Most midsize deals will not cross the thresholds. The risk is not that every acquisition is now caught. The risk is that an acquisition-led business assumes it is fine because each individual deal is small, and misses the cumulative trigger. Use the prompts below as a starting point for a conversation with competition counsel, not as legal advice.
Do You Need to Look Harder?
The honest answer for many readers is "probably not notifiable, but check the cumulative position." That check is cheap. Getting it wrong, and completing an acquisition that turns out to have been notifiable, is not.
How this fits a broader AI governance posture
Using AI in diligence is not a standalone decision. It sits inside how your organisation governs AI generally: who approves a use case, what data it can touch, who signs off on its outputs, and how you evidence all of that to a board or a counterparty. If you have not put that structure in place, the AI governance framework for Australian midsize business is the place to start, and the multi-entity financial consolidation guide covers the financial-diligence side for acquirers holding multiple entities.
The pattern across all of these is the same. AI is a leverage tool for expert humans working under real legal constraints. It reads faster, tracks more consistently and drafts more quickly than a tired team at the end of a long diligence period. It does not carry accountability, it does not understand privilege, and it does not decide whether you file. Treat it as the fastest junior analyst you have ever had, one whose work you always check, and it earns its place in the deal room. Treat it as an oracle, and it becomes the reason your filing is wrong.
What to do this quarter
- Build a three-year acquisition register. If you are acquisitive, maintain a live record of what you have bought, the Australian turnover involved, and the overlap categories. This is the single most useful thing you can do, with or without AI.
- Choose a private AI workspace for diligence. No deal data goes into consumer AI tools. Confirm data handling, retention and training settings contractually before anything sensitive is loaded.
- Write the sign-off rule down. Every AI-generated flag or draft is verified by a named human before it informs a decision or reaches a filing.
- Brief competition counsel early. The thresholds are cumulative and the regime is suspensory. Early advice is cheaper than a blown timetable.
- Fold it into your AI governance. Diligence AI is one use case among many. Govern it the same way you govern the rest.
The businesses that will move fastest under the new regime are not the ones that avoid AI, and they are not the ones that trust it blindly. They are the ones that use it to do the reading and the tracking, while keeping every legal judgement firmly in human hands.
Related Reading:
- AI contract review for midsize legal and procurement teams - The document-review capability applied to deal timetables.
- AI multi-entity financial consolidation for the midsize CFO - Financial diligence for acquirers holding multiple entities.
- ACCC consumer law and AI for Australian midsize business - Why "the AI said so" is not a defence.
- How the wrong AI tools leak business data - The confidentiality failure mode to avoid in a data room.
- AI governance framework for Australian midsize business - The structure that diligence AI sits inside.
Sources: Competition and Consumer Act 2010 as amended by the 2024 merger reform legislation; Australian Competition and Consumer Commission guidance on mergers and acquisitions; competition law analysis from Australian firms on notification thresholds and commencement (Gilbert + Tobin, Norton Rose Fulbright, Minter Ellison, DMAW Lawyers); Treasury announcements on transitional dates and 1 April 2026 asset and voting power triggers; Privacy Act 1988 and OAIC guidance on personal information and AI. Solve8 synthesis informed by enterprise integration experience across Australian organisations. This article is general information, not legal advice.