We have seen an increasing number of clients using Artificial intelligence (AI) to draft documents, correspondence or as a replacement of legal advice. The use of AI in legal settings can create substantial risks for clients which we wish to outline in this article. Because legal decisions can have serious financial, personal and professional consequences, relying on AI without appropriate human oversight can result in errors that are difficult to identify and potentially costly to correct.

‍

Accuracy and “Hallucinations”

One of the most significant risks associated with AI in legal work is that AI systems can produce information that appears convincing but is incorrect. This is sometimes referred to as an AI “hallucination”.

An AI system may invent a case, legislation, quotation, legal principle or citation that does not exist. It may also misinterpret an actual case or legislation, provide an outdated version of the law, or apply a legal principle to circumstances where it does not properly apply.

This is also exasperated by the fact that AI solely relies on the information you tell it. Clients often leave details out, or put greater emphasis on a factor which in their eyes is more important, but in reality a lawyer may consider less significant. It is unable to ask follow up questions or identify missing information.

This creates a particular problem in legal work because AI-generated material can appear authoritative. A person who is unfamiliar with legal research may not immediately recognise that an answer is wrong.

For example, an AI system might identify a court decision that appears directly relevant to a person's dispute. If the decision is incorrectly cited or the court's reasoning has been misunderstood or summarised, relying on that information could result in an individual making a poor legal decision or submitting inaccurate material to a court.

‍

The Law Changes

Another important risk is that AI systems may not always have access to the most current version of the law.

Legislation is regularly amended, regulations change, and courts continue to develop legal principles through new decisions. A response that was accurate at one point in time may become inaccurate following a legislative amendment or significant court decision.

This is particularly important in areas such as taxation, employment, immigration, corporate regulation and family law, where rules and policies can change over time.Clients should therefore be cautious about treating an AI response as a definitive statement of the current law. Legal information should be checked against authoritative and current sources.

‍

AI Does Not Replace Legal Judgment

Legal matters rarely involve simply finding a rule and applying it to a set of facts. Lawyers must often consider competing arguments, assess evidence, identify risks and determine how different areas of law interact.

AI can assist with these tasks, but it does not have professional responsibility for the consequences of its recommendations.

Two people can present apparently similar legal problems but require very different approaches because of differences in their evidence, objectives, contractual arrangements or personal circumstances.

A lawyer can ask follow-up questions, investigate the underlying facts and exercise professional judgment about how to proceed. An AI system may instead provide an answer based on incomplete or misunderstood information.

‍

Confidentiality and Privacy

Legal matters frequently involve highly sensitive information. This can include financial records, medical information, personal correspondence, business information, passwords, identification documents and details about disputes or criminal allegations. Providing such information to an AI system can create privacy and confidentiality concerns.

Clients should understand how an AI service stores, processes and uses information before entering confidential material. Particular care should be taken when dealing with information protected by legal professional privilege or confidentiality obligations, as it may later be deemed that you have waived legal professional privilege in circumstances you did not intend to.

‍

Bias and Incomplete Information

AI systems learn from large quantities of information, but the information available to them may contain errors, omissions or biases.

As a result, an AI-generated legal analysis may unintentionally emphasise some perspectives while overlooking others. This can be particularly problematic when dealing with complex disputes or matters involving competing factual accounts.

AI may also fail to appreciate information that has not been provided to it. A legal conclusion based on incomplete facts can be misleading even where the underlying legal information is technically correct.

‍

The Risk of Over-Reliance

Perhaps the greatest danger is not using AI itself, but relying on it too heavily. AI can be extremely useful as a starting point. It can help a person understand terminology, identify questions to ask a lawyer, or organise information to provide to a lawyer.

The problem arises when clients treats an AI-generated response as a substitute for professional legal advice. For example, someone facing a contractual dispute might ask an AI system whether they have a valid claim and then decide not to obtain legal advice because AI suggests that their position is strong. If the AI has overlooked an important contractual clause or procedural requirement, the person's decision could have serious consequences.

In the matter of Sadnan Khan v Aldi Pty Ltd [2026] FWC 3144, the Applicant used AI to file an unfair dismissal claim against his former employer after being terminated 3 days prior to the 6-month minimum employment period for protection from unfair dismissal.

After filing his claim, the Commission emailed him to advise that he appeared not to have a claim as he did not meet the minimum employment period, the Applicant continued his claim. The matter was then referred to the Deputy President to review, who then wrote to the Applicant explaining that he was unlikely to be successful. The Applicant continued with his claim. At the Hearing the Commission made a cost order against the Applicant in favour of the Applicant’s employer, and was ordered to pay for the employers legal costs for preparing and attending the hearing.

‍

A Better Approach to Using AI for Legal Matters

AI can still be valuable when used appropriately. Rather than asking AI to make the final legal decision, clients can use it to:

• explain unfamiliar legal terminology;

• summarise information that has already been verified;

• help organise documents and factual information; and

• identify questions that may need to be discussed with a lawyer.

The important distinction is between using AI as an assistance tool and treating AI as a legal authority or legal advice replacement.

‍

Conclusion

Artificial intelligence has considerable potential to improve access to legal information. However, its apparent confidence and ability to produce sophisticated language can make its limitations difficult to recognise.

Incorrect legal authorities, outdated information, privacy concerns, incomplete facts, bias and over-reliance are all significant risks. These risks become particularly serious when AI-generated information is used to make decisions involving substantial financial, personal or legal consequences, and interfere with a person from seeking legal advice where necessary.

AI can be a useful tool for understanding and organising legal information, but it should not automatically be treated as a substitute for professional legal advice. The safest approach is to use AI to assist human decision-making while ensuring that important legal information, conclusions and documents are independently verified.

‍

Need legal advice? Contact us to arrange a consultation with one of our experienced and dedicated lawyers on 07 3236 3999 or email us at reception@gsrlawyers.com.au.

‍