Unfair Contract Terms: An AI Audit Guide

Since 9 November 2023, using or relying on an unfair term in a standard form contract covered by the Australian Consumer Law has been illegal, and each unfair term is a separate contravention with corporate penalties up to $50 million. An AI-assisted audit finds every standard form template you use, maps risky clauses to known categories, and hands a lawyer a shortlist to judge.

What changed in the unfair contract terms regime?

For years the unfair contract terms (UCT) provisions had almost no downside. A court could declare a term void, but there was no penalty for including it, so many organisations updated their templates rarely and reviewed them less.

From 9 November 2023, using, applying or relying on an unfair term in a covered standard form contract became illegal and penalty-bearing under the Competition and Consumer Act 2010 and the ASIC Act 2001. Each unfair term is a separate contravention. A template with several problematic clauses is several exposures.

For a corporation, the maximum penalty is the greater of $50 million, three times the benefit obtained, or, where the benefit cannot be determined, 30 per cent of adjusted turnover during the breach period. The ACCC and ASIC have both named unfair contract terms as an enforcement priority.

What changed on 9 November 2023

• Unfair terms in standard form contracts became illegal and penalty-bearing, as well as void.
• Each unfair term is a separate contravention, so exposure stacks across a template.
• Corporate penalties reach the greater of $50 million, three times the benefit, or 30 per cent of adjusted turnover.
• Protection extends to contracts where a party has fewer than 100 employees or under $10 million turnover.
• The old upfront price threshold was removed, widening the regime's reach.

Who does the unfair contract terms law protect?

The regime covers standard form consumer contracts and standard form contracts where at least one party employs fewer than 100 people or has annual turnover under $10 million. The old upfront-price threshold, which many organisations relied on to argue a contract fell outside the rules, was removed.

For a mid-sized organisation, that matters in both directions. The terms you issue to smaller customers and subcontractors are likely in scope, and so are the standard terms your larger suppliers impose on some of your smaller counterparties.

Checking the threshold for one counterparty is easy. Checking it across thousands of customers and suppliers, and keeping it current as they grow or shrink, is a classification problem at a scale where manual assessment fails.

There is also a timing trap. The prohibition applies to contracts entered, renewed or varied on or after 9 November 2023. An old template that is renewed or has a term varied is squarely in scope, so an audit must reach the back catalogue still being renewed.

Why is a UCT audit a document-scale problem?

The three-part test is well established. A term is unfair if it would cause a significant imbalance in the parties' rights and obligations, is not reasonably necessary to protect the legitimate interests of the party it advantages, and would cause detriment if relied on.

The hard part is scale and drift. An organisation has services agreements, supply agreements, website terms, licences, equipment hire forms, subscription terms, and several versions of each because a template was forked for a big client years ago. Terms that were unremarkable before November 2023 may now be contraventions. Nobody made them unfair; they aged into a changed regime.

	Manual Review	AI-Assisted Review	Impact
Coverage	Sampled documents	Every template found	Complete
Consistency	Varies by reviewer	Same rules every time	Uniform
Clause mapping	From memory	Against a defined checklist	Traceable
Turnaround	Weeks per batch	Hours to a shortlist	Faster
Final judgement	Lawyer	Lawyer, on a shortlist	Focused

AI changes what the lawyer is handed. Instead of a filing cabinet, they receive a mapped shortlist with clause location and wording already extracted.

Where does AI help with standard form contract review, and where can it not?

Whether a clause is unfair depends on the whole contract, the commercial context and the legitimate interests of the party relying on it. No model should make that call, and no organisation should redraft contracts on an AI's say-so without legal review.

AI does three jobs well underneath that judgement.

1. Discovery. Contracts sit across drives, email, a contract system nobody fully trusts and personal folders. AI-assisted classification identifies which files are contracts, which are standard form, and which are near-duplicates of the same template. The same discipline drives supplier onboarding automation.
2. Clause extraction and mapping. Given a checklist of clause types that draw regulatory attention, a model locates matching clauses, pulls the exact wording and tags each against the checklist. See our guide to AI contract review that extracts key terms and risks.
3. Drift detection. After remediation, templates drift again as account managers fork them for deals. Comparing each executed contract with the approved template and flagging material deviations keeps the clean-up from unravelling.

1. 1DiscoverFind every template
2. 2ClassifyStandard form vs negotiated
3. 3ExtractPull clauses to a checklist
4. 4ReviewLawyer judges the shortlist
5. 5MonitorFlag future drift

Everything before review is preparation. Everything from review onwards is judgement. The same principle runs through procurement and legal ops contract review.

Which clauses draw regulator attention?

Regulators and courts have repeatedly flagged these categories, which give an audit its checklist:

• Broad unilateral variation rights letting one party change terms, price or service without agreement.
• One-sided termination powers that let the advantaged party walk away while binding the other.
• Automatic renewal with a narrow or impractical exit window.
• Liability for the other party's default, such as making the customer liable for the business's own failures.
• Disproportionate liability, indemnity and limitation clauses pushing risk onto the weaker party.

None of these is unlawful by definition. A variation right may be narrow and reasonably necessary. A limitation of liability may be balanced and standard for the industry. The categories are a map of where to look.

AI answers the descriptive question: do we have automatic renewal clauses, in which fourteen contracts, worded how? A person who understands the test answers the evaluative one: is this clause unfair here?

• Every standard form template located, including forgotten onesComplete coverage
• Flagged clauses mapped to known problem categoriesFaster legal review
• A defensible record of what was reviewed and whenEvidence of diligence
• Ongoing drift detection on executed contractsDurable compliance

If a regulator asks what steps you took to comply, a dated audit trail showing you found your templates, mapped them and had them reviewed is a far stronger position than an assurance that someone looked at some point.

How should a UCT audit be sequenced?

Cover the highest-exposure contracts first and build the ongoing control as you go.

1. Weeks 1-2InventoryDiscover and classify every standard form contract in use
2. Weeks 3-4PrioritiseRank by volume and value, starting with your most-used templates
3. Weeks 5-7Extract and reviewAI maps clauses to categories, lawyers judge the shortlist
4. OngoingMonitor driftCompare executed contracts against approved templates

A template issued to hundreds of smaller customers carries far more stacked exposure than a bespoke agreement used once, because the same term repeated across many contracts multiplies the contravention count.

What describes your contract portfolio?

• High-volume customer terms and conditionsStart with your most-used template
• Many supplier and services agreementsStart with supplier-facing standard forms
• Templates forked repeatedly for dealsStart with drift detection on executed contracts
• Unsure what templates even existStart with discovery and classification

What governance does an AI contract audit need?

Any system that reads your contracts handles commercially sensitive information and often personal information under the Privacy Act 1988. Know where documents are processed and stored, restrict who sees the output, and make sure no contract is amended or approved without a named person deciding.

The model never renders the legal verdict. An audit that ends with a lawyer reviewing a mapped shortlist is defensible. A script silently rewriting clauses is a different kind of liability. Our AI governance framework guide sets out the same guardrails.

The UCT regime sits beside the consumer guarantees the ACCC also polices. See ACCC consumer guarantees and AI implementation.

How Solve8 can help

We build contract discovery and clause extraction pipelines that run on your infrastructure in Australia, with your lawyers making every decision on what changes. See our custom AI development service and professional services work, or talk to us about auditing your templates.

This article is general information, not legal advice. Confirm your obligations with qualified legal advisers.