Will Artificial Intelligence Take Over Property Lawyers?

A comparative evaluation of the opportunities and limitations of generative artificial intelligence in Australian property law practice 

By Brad Marland, Marsella Zraika, Gorkem Oksuz and Mark Monfort 

Generative artificial intelligence (‘Gen AI’) is a form of artificial intelligence capable of producing new content by drawing upon patterns within the material on which it has been trained.¹ Its increasing adoption in property law has created opportunities in contract review, interpreting title searches and due diligence. However, its effectiveness must be evaluated against traditional practitioner review, with consideration given to the differing demands of residential and commercial transactions as well as the regulatory approaches adopted across different jurisdictions. 

Property law extends beyond identifying contractual terms and registered interests. Practitioners must interpret their legal effect and assess whether they restrict the client’s intended use of the property. Gen AI can rapidly summarise lengthy legal documents, compare clauses and flag issues of concern. These outputs can shorten the first stage of a review and support more detailed analysis. However, their usefulness depends on the information provided, the capabilities of the system and the practitioner’s assessment of their implications for the client’s intended use of the property. 

The pace of adoption is significant. Thomson Reuters reported that the proportion of legal organisations incorporating Gen AI into their work increased from 14% in 2024 to 26% in 2025. Its 2026 research also reported that 80% of legal professionals who use Gen AI use it for legal research. These are survey findings about the profession, not measurements of accuracy in property transactions. They nevertheless demonstrate why the question of effective supervision is increasingly practical rather than theoretical. [Sources: Thomson Reuters, “From Incubation to Integration” (2025); Thomson Reuters, “2026 AI in Professional Services Report” (7 April 2026). 

Gen AI and Traditional Review Compared 

Traditional practitioner review is thorough, but time-consuming. It requires the examination of each document individually and relies on the practitioner’s experience to identify unusual terms, inconsistencies or potential legal risks. 

In New South Wales, a contract for the sale of residential land must be accompanied by a series of prescribed documents, including title and planning documents,² among them a planning certificate issued under the Environmental Planning and Assessment Act 1979 (NSW).³ Gen AI is particularly effective in automating these administrative tasks by checking whether the required documents have been annexed, extracting zoning information from planning certificates and identifying discrepancies between the lot and plan numbers recorded in the contract and title search. 

These tasks are largely mechanical and repetitive, making them well suited to automation and potentially reducing the risk of human error. When appropriately configured and validated, AI-assisted workflows may reduce manual effort and help identify omissions or inconsistencies, although practitioners must still confirm the completeness and accuracy of the review. 

The distinction becomes more apparent when legal interpretation is required. Section 88B of the Conveyancing Act 1919 (NSW) permits the creation of easements, restrictions on the use of land and positive covenants, through instruments lodged with a deposited plan.⁴ Consider a drainage easement along the rear boundary of a residential property. While Gen AI may accurately identify the easement, it cannot reliably determine whether a proposed swimming pool would interfere with it or whether modification or extinguishment is possible under s 89 of the Act.⁵ Such an assessment requires consideration of the registered instrument, survey plans and relevant statutory requirements, in light of the client’s intended use of the property. This demonstrates the limitations of Gen AI in matters requiring legal interpretation rather than information extraction. 

Gen AI within Residential and Commercial Transactions

The value of Gen AI varies considerably according to the nature of the transaction. Residential conveyancing is generally standardised, with most contracts adopting the Law Society’s standard form. Consequently, significant contractual risks often arise from special conditions. Gen AI can assist practitioners by quickly identifying unusual terms such as provisions granting the vendor early access or imposing extended settlement periods. 

Commercial property transactions are generally more complex and less standardised than residential conveyancing. This creates greater opportunities for Gen AI, but also increases the risks associated with its use. Commercial leases often contain interdependent provisions governing rent reviews, permitted use, assignment, make-good obligations and termination. For example, a fixed annual rent increase of three per cent, may initially appear favourable. However, this advantage may be offset by an outgoings clause requiring the tenant to contribute to land tax and capital works. Similarly, Gen AI may identify an option to renew without recognising that it must be exercised within a prescribed period. Failure to identify this requirement could result in the tenant losing the right to renew. Further complexity arises in retail leases, where demolition clauses must be considered alongside statutory protections under the Retail Leases Act 1994 (NSW).⁶

In each of these situations, Gen AI may well flag the relevant clause, while failing to appreciate its relationship with the other provisions of the lease or with the governing statute. The comparative point is that the efficiency gains offered by Gen AI increase with the volume of documentation, but so too does the cost of a connection that goes unnoticed.

Comparing New South Wales and Other Jurisdictions

Property lawyers who engage in cross jurisdictional transactions, such as Madison Marcus, must ensure that their AI agents, and their review of the outputs, are tailored to the legislative requirements. For example, although the retail leasing laws across states are substantively similar, there are minor differences that, if not picked up, can void the clause or can be a liability. This is attributable to the manner in which AI agents respond to user inputs. Consequently, if a practitioner fails to indicate that the applicable jurisdiction has changed, the agent may continue to reference NSW legislation, notwithstanding that the matter concerns a Victorian lease. This ultimately highlights the importance of the practitioner’s discretion in reviewing the inputs, having a holistic understanding of the material and the law, and reviewing the final output.

The various jurisdictions have also adopted different approaches to Gen AI in court proceedings. The Supreme Court of New South Wales, adopted a prescriptive model in Practice Note SC Gen 23, restricting the use of Gen AI in the preparation of evidentiary material such as affidavits and expert reports and requiring that authorities cited in written submissions, be independently verified.⁷ The Supreme Court of Victoria’s Practice Note SC Gen 25, introduced in May 2026, similarly emphasises professional responsibility, verification, confidentiality and human oversight, while permitting AI-assisted preparation of court documents subject to its requirements. Although these practice notes concern litigation rather than ordinary conveyancing, they illustrate the importance of understanding the requirements applicable to each jurisdiction and context.⁸ 

Reliability and the Risk of Legal Error

The reliability of Gen AI remains the most significant constraint upon its use, because generative systems are capable of producing inaccurate information while presenting it with every appearance of certainty.⁹ That risk can no longer be described as hypothetical. This concern is illustrated in Handa & Mallick, where an Australian practitioner tendered a list of non-existent authorities generated using AI-assisted legal software.¹⁰ 

Although the case arose in litigation, its implications extend to property law, where errors may remain undetected until settlement or until a proposed development is refused. If Gen AI misidentifies a registered dealing, overlooks an easement condition, or mistakes the consequences of a lease option, the error can affect the overall outcome. The concern is not necessarily that Gen AI is more prone to error than a practitioner, but that its outputs can appear authoritative despite being inaccurate. This creates a risk that practitioners may accept incorrect information without sufficient scrutiny. 

Responsible use also requires attention to client confidentiality, legal professional privilege and data security. Before uploading contracts, title documents or other client material, practitioners should understand how the selected system stores, uses and shares that information, including whether it may be used for model training. Firm-approved tools, appropriate access controls and clear internal procedures are therefore important safeguards alongside verification of the output. 

These risks also call into question the efficiency gains associated with Gen AI. Time saved through automated review may be offset by the need to independently verify its findings. Where errors remain undetected, the resulting legal and financial consequences may outweigh any initial savings. Ultimately, efficiency cannot come at the expense of accurate legal advice and the protection of clients’ proprietary interests. 

Conclusion

Gen AI offers genuine advantages for property lawyers in organising documentation and generating preliminary issue lists. Those advantages apply to both residential and commercial property work, although commercial property transactions often demand closer analysis of interacting contractual provisions. Adoption statistics indicate growing professional use, but they do not establish that AI can independently deliver reliable property advice. The central requirement remains that practitioners verify the information identified and exercise independent judgement when assessing its legal and commercial significance.

Gen AI is therefore a valuable tool to aid property law practitioners by improving efficiency and assisting with the identification and review of legal information. However, its effectiveness remains dependent on independent legal judgment and professional oversight. Its value lies not in replacing practitioners but in supporting their work, while maintaining the accuracy of legal advice and protecting clients’ proprietary interests. Ultimately, successful adoption should be measured by whether Gen AI improves the quality and efficiency of property law services without compromising the interests of the client. 

Our experienced team of property lawyers at Madison Marcus are at the centre of the use of AI in streamlining process, while using their industry experience to validate and ensure our clients receive the highest level of advice and attention.