.svg.png?noresize&width=1920&height=960&name=Flag_of_Australia_(converted).svg.png)
Australia
Tools
Tasks
Users
Scope
Training
Regulation
Cases
Insights
Information uploaded as at July 2026
AT A GLANCE
In Australia, AI use in criminal proceedings is limited, and tends to be concentrated at the investigative stage, with little to no engagement at the court stage. Law enforcement agencies are the primary users of AI, deploying tools principally for data review and case management. Legal practitioners have adopted AI for preparatory tasks such as document review, summarisation, and translation, but not in any active role during proceedings, despite compelling uptake statistics in a recent Victorian lawyers census. Based on publicly available information, courts have largely avoided the use of AI tools in criminal proceedings. There is no dedicated regulatory framework governing the use of AI in criminal proceedings, and no mandatory or systematic training for legal professionals on AI.
Australia has no single law governing AI in criminal proceedings; regulation instead operates through a fragmented patchwork of court practice notes, judicial and professional body guidance, and general law statutes (privacy, copyright, consumer protection), anchored loosely by a non-binding federal AI policy. Coverage is uneven across jurisdictions, with judicial guidance issued in only a few states and the Federal Court. Australian courts have now encountered more than 30 cases involving AI-generated fabrications submitted to the court, with consequences ranging from costs orders to professional disciplinary referral. A national court-centred framework is reportedly in development, but none has yet been published.
USE
Australia is a federation, comprising the Commonwealth Government, six States and two self-governing territories, each with their own laws, courts and approaches to criminal justice. The AI Justice Atlas does not attempt to document developments in every jurisdiction; rather, it offers a high-level overview of how AI is being integrated into criminal proceedings, and highlights significant initiatives and patterns across the broad legal landscape.
The goals driving AI use in criminal proceedings in Australia vary by institution but centre on saving time and reducing costs. At the same time, official sources have identified a number of shared concerns over the use of AI tools in criminal proceedings, including: a lack of transparency; insufficient awareness of AI's limitations; and concerns about accountability.
Law enforcement
Australian police forces across federal, state, and territory jurisdictions have adopted a wide and growing range of AI tools, some of which are present in more than one jurisdiction. Most of these tools serve an operational support function rather than replacing investigative judgement. Given the breadth of this landscape, the information below focuses on selected major tools.
Operational support
The Australian Federal Police has also publicly disclosed the six types of operational AI tools that make up its AI capability. All outputs generated by these systems require—based on the AI Transparency Statement of the Australian Federal Police—human review before they can influence any operational decision affecting individuals. As at July 2026, no further details on these tools have been made public and the Statement has not disclosed the specific AI tools that it refers to:
|
Capability |
AI Type |
Status |
|
Audio Transcription |
Natural Language Processing (NLP) |
In production, operational use |
|
Language Translation |
Natural Language Processing (NLP) |
In production, operational use |
|
Optical Character Recognition (OCR) |
Computer Vision |
In production, operational use |
|
Object Recognition |
Computer Vision |
In production, limited operational use |
|
Autonomous 3D Mapping |
Computer Vision |
In production, limited operational use |
|
Large Language Model (text-to-text) |
Natural Language Processing (NLP) |
In production, limited general use, jurisdictional restriction applies |
In 2025, it was reported that Victoria Police integrated generative AI into its Police Assistance Line and Online Reporting platform to automate the processing of non-urgent incident reports. The system relies on the Salesforce OmniStudio AI tool to manage around 20% of crime reports, using it to generate summaries of community-submitted online reports for officers, though the contents are still handled by a human agent before referral to frontline officers. By replacing the manual editing procedures previously managed at central police stations, this initiative ensures that information is delivered to officers in a familiar, professional format while significantly increasing administrative efficiency.
Predictive analytics
The Queensland Police Service has trialled an AI tool, which it announced in 2021, to improve responses to domestic and family violence. Its purpose is to predict and prevent serious domestic and family violence incidents before they occur, by analysing large volumes of historical police data to identify individuals assessed as high-risk offenders and rank them according to their likelihood of causing harm. Using this risk-assessment, the system enables a form of preventative policing in which officers can proactively intervene, for example by visiting individuals flagged by the algorithm, rather than waiting for emergency calls or incidents to escalate.
Data review and analysis
Various police forces in Australia are using AI for facial-recognition capabilities:
|
NEC’s NeoFace |
NEC’s NeoFace is a facial recognition tool designed to assist with suspect identification and reduce investigation times for cases involving facial video evidence. It achieves this through both 1:1 image comparisons and broad searches against large databases. The system draws from various sources, including institutional and police databases, crime scene CCTV, document scans, surveillance imagery, and social media. NEC’s NeoFace produces a ranked candidate list for investigator review, but it is treated strictly as a discovery tool rather than standalone evidence. As at July 2026, the system is being operated by the Australian Federal Police, South Australia Police, and the Northern Territory Police. |
|
iFace |
Victorian Police have also made use of the AI-powered iFace tool, which functions as a facial recognition technology system used in investigations to identify criminal suspects. Victoria Police uses iFace to search for persons of interest in an offender image database. In 2020, there were 86 police stations across Victoria with iFace capture points, which allowed offenders’ facial images to be captured and then added to the image database. |
In 2024, it was reported that Queensland Police is trialling Truleo AI, as part of efforts to address domestic violence through improved analysis of officer interactions. The tool reviews and interprets body-worn camera footage using AI, converting spoken interactions into structured data that can be analysed to assess how officers handle sensitive situations, identify patterns, and highlight best practices or areas for improvement. By applying natural language processing and machine learning techniques, it evaluates communication in real-time or retrospectively, aiming to support better decision-making and training within police departments rather than predicting criminal behaviour directly.
Acusensus一a private company that collaborates with governments and commercial stakeholders to tackle distracted driving globally一has provided various Australian law enforcement authorities with an AI camera programme, which detects illegal mobile phone use while driving, as well as non-compliance with seatbelt laws. Each camera relies on two HD cameras and an IR flash. AI software screens every passing vehicle, and automatically filters out non-offending images (deleted within approximately one hour), with potential offences verified by authorised human reviewers before fines are issued. As at July 2026, Acusensus cameras are running in New South Wales, Queensland, South Australia, Western Australia, and the Australian Capital Territory.
Western Australia Police have reportedly piloted a cloud-based platform developed by Modis, powered by Microsoft Azure Cognitive Services, to process large volumes of evidential data (emails, messages, social-media posts and footage). It was reported that the platform identified 18 new points of interest within 24 hours in one investigation.
New South Wales Police and South Australia Police use Auror, a retail crime intelligence platform that combines AI-assisted offender profiling, facial recognition, and automatic plate recognition data to link retail crime incidents and identify repeat offenders across jurisdictions.
Since February 2021, the Police of New South Wales has used its Insights platform, enhanced by Microsoft Azure Cognitive Services, to manage vast quantities of digital evidence. By leveraging computer vision and machine-learning, the system reviews a broad range of sources (including police and public CCTV footage) to automatically identify objects, vehicles, and specific visual features within CCTV, body-worn cameras, and dashcam footage. Additionally, the platform uses voice-to-text technology to transcribe police interviews, processing approximately one terabyte of computer vision data every day. The system analyses a broad range of sources including police and public CCTV footage. Critics have argued that the platform’s reliance on location data disproportionately implicated innocent people in high-crime areas.
The Australian Federal Police and Monash University’s AiLECS Lab have collaborated to produce Silverer, an AI-disruption tool designed to combat the creation of harmful deepfakes. By applying subtle, ‘pixel-level’ changes to images (known as data poisoning), Silverer ensures that any AI models trained on that data produce distorted, unrecognisable results rather than realistic media. This process is specifically intended to prevent the generation of child abuse material, extremist propaganda, and unauthorised deepfake videos.
Finally, in October 2025, the Australian Federal Police announced that they were developing, with Microsoft, a prototype tool to interpret emojis and GenZ and Alpha generation slang in encrypted communications used by online predators, to identify harmful intent earlier and protect children.
Prosecutors
As at July 2026, there is no uniform use of AI by prosecutors across Australian jurisdictions. While most states and territories have established positions on AI in prosecutorial activities (see below), few have acquired or developed concrete tools, with the exception of New South Wales and Victoria Offices of Public Prosecutions.
Case management
Between 2022 and 2023, the New South Wales Office of the Director of Public Prosecutions replaced its legacy, on-premises management system with a new platform built on Appian, a low-code process automation environment. The office rebuilt the system from scratch, replicating all existing functionality while migrating 20 years of case data. This new platform offers improved remote access across metropolitan and regional offices, a simplified interface, and enhanced automation that reduces manual, paper-based tasks. The platform integrates directly with NSW Forensic and Analytical Science Services and the JusticeLink court management system — respectively, a branch of New South Wales Health providing independent forensic analysis in support of the New South Wales justice system, and the electronic case management system used by New South Wales courts and tribunals.
Similarly, the Office of Public Prosecutions Victoria developed Amicus in 2022. Also built on Appian, Amicus is designed to replace the Office’s legacy case and document management systems (some over 30 years old) that were primarily paper-based and unable to meet modern demands. Amicus integrates previously siloed systems into a single, unified interface, providing stakeholders with a consolidated, real-time view of data. As at July 2026, Amicus features a Record of Decision Builder, enabling prosecutors to request approvals from Crown Prosecutors via integrated workflows. Further enhancements planned for the platform include a Police Portal for evidence uploads, an application for victims for case updates, and a bespoke generative AI tool named Amicus X (see below). In 2024, Amicus won the 2024 Victorian Premier’s Design Award in the Design Strategy category.
Predictive analytics
The Victoria Office of Public Prosecutions’ Amicus platform (see above) features an Indictment Builder, which automates sentence calculations for complex matters.
Legal research, analysis and drafting support
Though not in operation as at July 2026, as noted above, the Victoria Office of Public Prosecutions is developing Amicus X to ingest police briefs and process them through an LLM to provide rapid, structured overviews and summaries. The tool is expected to generate detailed evidence charts cross-referenced with applicable legal provisions, produce interactive footnotes, and identify evidentiary inconsistencies. Furthermore, it seeks to flag potential gaps where evidence may be insufficient, to support the prosecution of specific offences.
Courts
Most Australian courts have adopted a notably cautious stance toward the use of generative AI. Some courts even restrict or prohibit its use for particular categories of material, unless strict conditions are met or the court grants leave (see below). Several Chief Justices of Australian courts,, including Chief Justice of the High Court Stephen Gageler AC, Chief Justice of New South Wales Andrew Bell AC, and Chief Justice of Queensland Helen Bowskill have publicly voiced concerns about the misuse of AI and have called for strict regulation.
Despite this overall cautious posture, some AI tools are being piloted and introduced. As at July 2026, most of these tools have been concentrated in Victoria, which has been experimenting with AI applications since early 2024. There is an emphasis on increasing the efficiency of justice, but also the need for responsible use of AI.
The Court recognises that Generative AI has the potential to facilitate the just resolution of disputes by increasing efficiency in the conduct of litigation, reducing legal costs, enhancing access to justice and the quality of the administration of justice. . . It is fundamental to the administration of justice that Generative AI must be used appropriately, responsibly and with due care. Otherwise, Generative AI poses risks to the proper administration of justice and public confidence in the legal system.
Case management
The Coroners Court of Victoria—an independent, specialised, and inquisitorial court that investigates reportable deaths and fires—has piloted AI-Powered Image Redaction, which automatically identifies and redacts distressing images, such as photographs of deceased persons or accident scenes, from case documents before staff review them. Its primary goal is to mitigate vicarious trauma among employees routinely exposed to traumatic content. The pilot also explored the ability of the AI system to scan large documents and surface key information, assisting coroners, solicitors, and investigators with their reviews.
The County Court of Victoria is trialling an AI transcription solution to replace manual transcription of hearings, which normally takes between one and fourteen days depending on length and urgency. The system allows transcripts to be reviewed in real-time during a hearing, and also enables practitioners and judges to cross-reference transcripts from earlier hearings in the same case while proceedings are ongoing. Strict security controls apply, including data sovereignty requirements that ensure all audio remains within Australia.
The Supreme Court of Victoria has occasionally used specialised third-party tools designed to assist with discovery. In the civil law case McConnell Dowell Constructors (Aust) Pty Ltd v Santam Ltd & Ors (No 1) [2016] VSC 734, the Court ordered the use of Technology-Assisted Review (‘TAR’), an AI-driven machine learning process used to classify vast document sets for relevance and privilege during civil discovery. Faced with around 1.5 million potentially relevant documents stemming from a natural gas pipeline insurance dispute, Justice Vickery calculated that a manual review would occupy a junior solicitor for over 583 working weeks (ten years of work). Consequently, the Court appointed a Special Referee and, with the parties’ cooperation, endorsed a TAR-based discovery protocol, marking the first time an Australian court had ordered predictive coding to ease the discovery process. Other courts have since followed a similar practice, including the Federal Court of Australia.
Briefcase offers an AI case management platform that reduces the time from filing to decision. It uses AI for managing, for example, party correspondence, real-time transcription and document assembly, but never the decision, which stays firmly with the judge. The platform is configurable to any jurisdiction or case type.
Legal research, analysis and drafting support
In its October 2025 Report, ‘Artificial Intelligence in Victoria’s Courts and Tribunals’, the Victorian Law Reform Commission recorded that the Law Library of Victoria is trialling LexisNexis and Westlaw AI products for possible future use.
Chief Justice Gageler AC has reported that the High Court of Australia planned to run a pilot in 2025 to test closed AI programmes for editing judgments, noting that such a tool would not generate the original text but improve it. As at July 2026, there are no reports about its performance or continued use.
Defence
As at July 2026, the use by defence lawyers in Australia is principally driven by the adoption of both general-purpose and law-specific commercial tools (used for legal research, drafting, and litigation-support) rather than by bespoke tools developed within the criminal defence sector itself.
In 2025, the Victorian Legal Services Board and Commissioner’s Lawyer Census report provided insights on the use of AI by lawyers within its jurisdiction. The Victorian Legal Services Board and Commissioner is an independent statutory authority whose role is to regulate the legal profession in Victoria. It found that over a third (36.7%) of Victorian lawyers surveyed were using AI in their practice, though actual use may be higher since lawyers may be unaware that they are using AI, including in common tools such as Google Search. The Census found that: (1) among those surveyed, over half reported regular use—weekly (35.1 per cent) or daily (25.2 per cent)—of AI; (2) public AI tools (such as Claude) are used more frequently than specialised legal tools (such as Lexis+AI); and (3) AI is used in ways that may interact with Victoria’s courts, such as in legal research and drafting court documents.
Administrative support
Nuix Neo Discover and RelativityOne are used for eDiscovery by lawyers in Australia:
|
Nuix Neo Discover |
Nuix Neo Discover is an AI-powered eDiscovery and legal review platform. It is powered by Nuix’s proprietary Cognitive AI to allow AI-enabled search, and over 500 pre-built eDiscovery-tuned models. The broader Nuix Neo engine can ingest and analyse more than 1,000 file types, allowing legal teams to process large and complex datasets for review. The platform uses AI-driven workflows to assist with document categorisation, predictive coding, concept clustering, semantic search, PII identification and redaction, and AI-assisted first-pass review for e.g. relevance, confidentiality, and privilege classifications. Nuix presents the platform as capable of identifying gaps and inconsistencies in document collections or productions, revealing hidden patterns, and presenting enriched data insights through enhanced visualisations. Nuix Neo Discovery also supports continuous active learning based on reviewer input. |
|
RelativityOne |
RelativityOne is another cloud-based eDiscovery and legal review platform that uses AI to help legal teams process, review, and extract insights from large volumes of digital evidence. Its AI capabilities are delivered through the Relativity aiR suite, which comprises three generative AI products built directly into the platform:
All three products provide document-level citations and natural-language reasoning. |
Mary Technology is also a task-specific tool, but it focuses on assisting users with case chronologies and fact timelines, built from large volumes of unstructured legal material. It can ingest raw case files, including medical notes, witness statements, correspondence, emails, audio files, and documents in more than 200 formats, and convert them into structured, source-linked, and searchable chronologies. It also highlights inconsistencies and evidential gaps. The system combines traditional machine learning with large language models, including ChatGPT. It also integrates with Smokeball, a legal practice management AI software for law firms.
Finally, Vici is an AI analytics system that allows users to interrogate large datasets quickly, uncover connections between documents, identify patterns, and support decision-making through interactive visualisations. The platform can process hundreds of file formats, including emails, audio files, text messages, and the full Microsoft Office suite, within seconds. It then presents results through timelines, maps, and webs showing relationships between entities and documents. Although Vici was not developed exclusively for the legal market, it can support litigation and discovery by identifying links and patterns across large volumes of digital evidence. Its Business Development Managers have described it as a tool built to find ‘that needle in the haystack’ and reveal connections that would otherwise take weeks to identify manually.
Legal research, analysis and drafting support
Australian lawyers, including defence counsel, are using a variety of tools for legal research, analysis and drafting support:
|
AI Legal Assistant |
AI Legal Assistant is a platform designed by practising lawyers for Australian and New Zealand legal practices. It offers AI-assisted legal research grounded in Australian and New Zealand law, with new cases added weekly. It also supports document analysis and review, identifying ambiguities, complex language, conflicting terms, inconsistencies, missing clauses, non-standard provisions, and negotiation points. In addition, it provides AI-assisted drafting of correspondence, pleadings, and other legal documents, as well as voice-to-text conversion. For litigation practitioners, including defence lawyers, the platform expressly supports the generation of defence pleadings and interlocutory applications, legal research with Australian caselaw citations, analysis of expert reports and technical evidence, and discovery document review. All data is stored on Australian servers. |
|
CoCounsel |
CoCounsel is another generative legal-specific AI platform integrated into Thomson Reuters's Westlaw and Practical Law databases. It offers multiple core functions, including preparing for depositions, drafting correspondence, searching databases, reviewing and summarising documents, extracting contract data, checking contract policy compliance, building timelines, and agentic guided workflows for multi-step legal tasks. It also includes a Deep Research (Agentic AI) function that plans, reasons across the full Westlaw content set, and delivers structured research reports with arguments on both sides of legal questions. All customer data is stored and processed in Australia, and Thomson Reuters does not use customer data to train its models. The tool is available across all areas of law, including criminal defence. |
|
CourtAid |
Australian sole practitioners and small criminal defence firms also use CourtAid, an AI legal research tool that enables them to engage conversationally with current legislation and court judgments. It covers more than two million judgments, statutes, and legislative instruments across Australia, the United Kingdom, and New Zealand, with data updated daily. The platform is trained on verified legal materials, reducing the hallucination risks associated with some other AI research tools in the Australian market. CourtAid is positioned as an affordable option, making it accessible to those who cannot afford enterprise-grade legal research platforms. |
|
Habeas |
Habeas is an AI-powered legal research and document analysis platform built specifically for the Australian legal system. It uses semantic search across a corpus of more than 300,000 Australian judgments and statutes, returning paragraph-specific results with clickable citations for verification. The system is ‘search-first’, which means that it generates legal reasoning only after checking it against actual legal authority, which the company presents as a structural safeguard against hallucination risks. The platform also allows lawyers to upload and query their own documents, and supports configurable AI research assistants tailored to specific practice areas. Habeas has also been integrated with the case management platform Smokeball. |
|
LawY |
LawY is an AI legal research assistant integrated directly into two of the largest legal practice management software platforms available in Australia: LEAP and Smokeball. It provides jurisdiction-specific answers to legal research questions, with linked citations, across all Australian jurisdictions and areas of law, including criminal law. It offers three levels of research: (1) Instant Answers for quick queries; (2) Rapid Research, and (3) Deep Research. Its most distinctive feature is its ‘lawyer-in-the-loop’ model, a human verification layer that allows users to submit selected AI-generated answers for review by a panel of experienced, qualified lawyers in the relevant jurisdiction and practice area. These lawyers check and correct the answers before they are returned. The tool also follows a privacy-by-design approach and does not use user inputs to train AI models. |
|
Lexis+AI |
Lexis+AI (or Protégé) is a generative, legal AI platform built on LexisNexis’s proprietary Australian legal content database, spanning over 1.23 million documents encompassing legislation, caselaw, and commentary. It delivers conversational legal research, enhanced document summarisation, intelligent legal drafting, and document upload and analysis. Responses are grounded in the LexisNexis content repository through retrieval-augmented generation, with linked citations provided to reduce the risk of fabricated content. The tool uses a multi-model approach, drawing on large language models including Claude and ChatGPT. The tool is available across all areas of law, including criminal defence. |
Victims
Under Australian criminal procedural rules, victims generally do not have independent standing in criminal proceedings. Victims may, however, participate as witnesses.
Case management
There are, however, a number of victim-facing AI tools available in Australia. For example:
|
Hear Me Out |
Hear Me Out, developed by the National Justice Project, is an AI-powered complaints and legal self-help platform. A user describes what happened; the system helps identify an appropriate complaint pathway. With an initial focus on New South Wales and Federal complaint bodies, the website provides a range of legal resources, factsheets and guidance about navigating more than 75 complaint bodies. In the first six months of the New South Wales pilot, more than 400 complaints were assisted through the platform. A third of issues raised related to police, legal, human rights, and discrimination complaints. Almost a quarter of complaints involved the public sector. Consumer rights and business disputes represented 16% of complaints. |
|
Umibot |
Umibot is an AI tool developed by RMIT University, specifically designed for victim-survivors of image-based sexual abuse. The tool provides information about Australian criminal and civil law, as well as guidance on reporting crimes to police and eSafety, collecting evidence, reporting options, support services, cybersecurity and online safety. |
TRAINING
As at July 2026, there is no official mandatory or systematic AI training for legal professionals in Australia. However, Australia has a multitude of various AI training initiatives for justice sector professionals, developed at the level of individual professional bodies, judicial education institutions, and research organisations. Activity is most advanced among lawyers, where law societies and bar associations across several jurisdictions have been delivering structured continuing professional development on AI since 2023. For judicial officers and tribunal members, a small number of dedicated training events have taken place at the national and state levels. For law enforcement, there are at least two substantive initiatives reported so far. However, for prosecutors, there have been no publicly reported AI training initiatives as at July 2026. There may be utility in developing prosecutor-specific AI training or, should such tailored training not be available, in engagement by prosecutors with AI training related to criminal law offered by law societies and bar associations.
For law enforcement, the most significant AI training activity is targeted to the police and consists of an AI Masterclass. The Masterclass is a joint initiative of the Australian Institute of Police Management and the Australian Federal Police. The Masterclass is designed to equip law enforcement leaders with the knowledge and skills to understand AI and navigate the complexities of public trust. Its stated learning objectives include:
- Increasing awareness of AI categories and types and their societal impact;
- Understanding the threat posed by AI, with real-world case studies on generative AI and child exploitation;
- Recognising the importance of AI for law enforcement and addressing the challenges of data volume and scalability; and
- Comprehending the implications of AI policy on law enforcement and the national intelligence community.
For judges, the National Judicial College of Australia (the principal body for judicial education across all Australian court hierarchies and jurisdictions) included in its National Curriculum, a dedicated unit entitled ‘Intersection of Science, Technology and the Judicial Function’, which includes a dedicated programme on ‘AI and the Courts’. The National Judicial College of Australia has also carried out specific events for Australian judges, together with the Australian National University, such as a Joint Conference on Artificial Intelligence and the Courts, held in Canberra on 6-7 April 2024. This two-day programme was designed to bring together the judiciary, leading academics from Australia and overseas, policymakers, and technical experts, with sessions led by experienced members of the judiciary.
The Federal Court of Australia has also issued internal guidelines, under which Court-wide training will ensure awareness of, and foster compliance with, defined policy and best practice in relation to the use of AI and data protection.
Beyond efforts of the National Judicial College and the Federal Court, several federal colleges and courts have advanced AI literacy initiatives, for example:
- The Judicial College of Victoria has developed an ongoing AI series for judicial officers. The series is designed to encourage judicial officers to engage with both the risks and potential of AI, and to learn about the implications for the judiciary in an era of rapid technological advance. It includes past and upcoming seminars on topics including deepfakes, AI and evidence, AI in the future of the judicial role, and AI prompt engineering.
- The Council of Australasian Tribunals New South Wales, on the other hand, organised two events in 2025 addressed specifically to tribunal members. A seminar series titled ‘AI Revolution in Law’ was held on 10 April 2025 at the Personal Injury Commission in Darlinghurst, discussing the practical applications and professional impacts of AI, with a particular focus on how AI affects the work of tribunals. On 21 May 2025, the Council hosted its annual Whitmore Lecture, titled ‘Artificial Intelligence and Tribunals’, delivered by Professor Tania Sourdin. The lecture explored how AI can enhance tribunal operations by examining supportive, replacement, and disruptive technologies, drawing on regulatory requirements, governance arrangements, and international AI developments to consider how courts and tribunals can adapt.
For lawyers, Australian federal bar associations and law societies have also organised development seminars on AI and its impact on the legal profession:
|
Bar Association of New South Wales |
The Bar Association of New South Wales has hosted seminars sessions on topics such as ‘AI, Robots and the Law’, and ‘AI, Technology and Human Rights’. |
|
Law Society of New South Wales |
The Law Society of New South Wales contributed to practitioners’ learning of AI through resources on responsible AI use, key AI terminology, court protocols, professional obligations, tool selection, hallucination risks, and risk mitigation. These resources include various on-demand courses, such as:
|
|
Law Society of the Australian Capital Territory |
The Law Society of the Australian Capital Territory has hosted a three-session event in March 2025, covering ‘AI Essentials for Sole Practitioners, Small Firms, and Legal Practices’; ‘Legal-Specific AI Tools’; and ‘Responsible Practice in 2025: Ethics, Risks, and Managing AI Challenges’. All sessions are available on-demand through its webinar library. The Law Society also offers continuing professional development seminars on various topics, including AI, and, as at July 2026, provides an on-demand course titled ‘From Automation to Intelligence: Understanding Evolving Technology in Administrative Law’. |
|
Law Institute of Victoria |
The Law Institute of Victoria runs events and webinars on AI through its AI Hub, including a Young Lawyers Technology Masterclass programme and a related technology series addressing the skills junior lawyers and law students may need when using AI. Additionally, the Law Library of Victoria (the library service of the Supreme Court of Victoria), runs free online training sessions for Victorian legal professionals that directly address AI. Programmes include: ‘Introduction to AI and How it is Used for Legal Research’ and ‘Decoding AI in Case Law’. |
|
The Piddington Society |
The Piddington Society (a Western Australian legal continuous professional development provider), delivered its ‘Piddington AI Program’ in March 2025, comprising three structured sessions recorded during the 2024–25 year. The programme covered how AI works, how to use AI in practice management and legal operations, and how to govern and manage AI in a legal workplace. It is available on demand. |
|
Australian Capital Territory Bar Association |
Moreover, the Australian Capital Territory Bar Association announced, in December 2025, a partnership with AI Legal Assistant to deliver formal AI training to Canberra barristers. Under the arrangement, AI Legal Assistant will run a series of education workshops on how to use generative AI tools safely and accurately in legal work, over 18 months from early 2026. |
REGULATION
As at July 2026, there are no statutes expressly regulating the use of AI in criminal proceedings in Australia.
The Australian Government has released a Policy for the Responsible Use of AI in Government, which applies to police, the courts, and prosecutors. The Policy provides that all non-corporate Commonwealth entities are bound to apply it, which relevantly includes the Federal Court of Australia and Australian Federal Police. The Federal Court, in its AI Transparency Statement, has stated that it is committed to the safe, ethical and responsible use of AI in accordance with the Policy. The Australian Federal Police announced in February 2025 that, in accordance with the Policy, they had appointed an accountable official, which is the individual accountable for implementing the Policy within the entity. However, any compliance by police with the Policy would be subject to the national security carveout provided for in the Policy, in relation to the intelligence role and functions of the Australian Federal Police.
Aside from the Policy, the AI-regulation landscape is generally defined by sector-specific guidance issued by relevant bodies, as well as the obligations imposed by general law statutes and the ethical duties of legal practitioners, which extend to ensuring the responsible and effective use of AI.
AI regulations
Australian Government, Policy for the Responsible Use of AI in Government, 2025
The 2025 Policy for Responsible Use of AI in Government aims to establish a framework facilitating a coordinated approach to the government’s use of AI. It is binding upon public actors in the justice sector (see above) but lacks any real means of enforcement as a soft policy instrument. However, actors’ accountable officials must share a register of in-scope AI use cases with the Digital Transformation Agency and notify the Agency if any high-risk use cases are identified. The relevant requirements it imposes upon actors are as follows:
|
Strategy and Oversight |
Agencies must make a publicly available (and annually reviewed) statement explaining their approach to the use of AI. They must also develop a strategic position on AI adoption and designate accountable officials responsible for implementing the Policy. |
|
Preparedness and Operations |
Agencies must create an approach to embedding responsible AI practices in their functions. Mandatory training for all staff on responsible AI use must also be implemented. |
|
AI Use Case Impact Assessment |
Agencies must assess and document all AI use cases against in-scope criteria. An AI use case is deemed to be ‘in-scope’ if, for example, its use or misuse could lead to more than insignificant harm or it is possible the public will directly interact with, or be significantly impacted, by the AI or its outputs without human oversight. |
Guidelines for practitioners
Guidelines for practitioners on the use of AI in criminal proceedings in Australia operate through a patchwork of rules. Courts, law societies, and bar associations in most, but not all, jurisdictions have issued separate guidance for judicial officers and for practitioners or litigants, generally distinguishing between AI as an aid to preparatory work and its exclusion from judicial decision-making or the generation of evidence.
Guidelines for judicial office-holders
As at July 2026, guidance on the use of AI by judges in Australia has been issued in New South Wales, Victoria, and Queensland, in addition to the Federal Court. No guidelines specifically addressing judicial use of AI have yet been issued in Western Australia, South Australia, the Australian Capital Territory, the Northern Territory, or Tasmania. Where guidance has been issued, courts consistently adopt the view that AI cannot replace judicial decision-making, but may support preliminary or preparatory tasks, particularly research, summarisation, and proofreading.
|
Court/Federation |
Policy |
Content |
|
Federal Court of Australia |
AI Transparency Statement (April 2026) |
The Federal Court’s AI Transparency Statement provides that the Court takes a ‘cautious approach’ by permitting the use of generative AI tools ‘within defined policy restrictions’. Thus, the Federal Court does not use AI ‘in any way where the public may directly interact with it or be significantly impacted by it’. Specifically, AI is not used for decision-making, data analytics, prediction, service delivery, or policy and legal activity. In addition, obligations attach to any use of AI, including requirements to log and monitor its use. With this in mind, the Court has issued internal policy guidance with the aim of addressing inherent risks associated with the deployment and use of generative AI tools, while recognising the opportunities that such tools present. The policy controls and restricts the use of generative AI tools, anticipates logging and monitoring of their use, articulates technical and ethical risks that must be considered when using such tools, and requires that any use of permitted AI tools is consistent with the Court’s policies. |
|
New South Wales |
Guidelines for New South Wales Judges in respect of the Use of Generative AI (November 2024) |
In addition to Practice Note SC Gen 23 (3 February 2025) (discussed below), the Chief Justice of New South Wales issued separate Guidelines for New South Wales Judges in respect of the Use of Generative AI. The Guidelines caution judges that prompts or interactions with a generative AI chatbot may, unless disabled, be added to the underlying large language model database and used to respond to other users’ queries, raising confidentiality risks given the sensitive material judges may deal with. The Guidelines also flag the possibility that training data underlying a generative AI tool may have been obtained in breach of copyright, and the risk that granting an AI tool device ‘permissions’ may inadvertently expose information stored on a judge’s or judicial staff member’s personal devices. Judges are directed that any output from generative AI research must be independently verified before being relied upon. |
|
Queensland |
Use of Generative AI Guidelines for Judicial Officers (September 2025) |
Queensland Courts have issued ‘Use of Generative AI Guidelines for Judicial Officers’, applying to all courts in Queensland, the Queensland Industrial Relations Commission, and the Queensland Civil and Administrative Tribunal. The Guidelines do not impose a ban on the use of AI in relation to ‘research or preparatory’ work, and the Guidelines remind judicial officeholders to check the accuracy of information provided by the AI tool and to maintain best practice in ensuring the secure use of AI. The Guidelines also note that judicial officeholders hold personal responsibility for the material produced in the judicial officer’s name. But the Guidelines also provide that generative AI could be used as a ‘potentially useful secondary tool for research or preparatory work’, although it ‘should not be used for decision-making nor used to develop or prepare reasons for decisions’. |
|
Victoria |
Guidelines: The Use of Artificial Intelligence by Judicial Officers (May 2026) |
The Supreme Court of Victoria was among the first Australian courts to issue AI guidance for judicial officers, in 2024, and updated this guidance in May 2026 with a new Practice Note (for court users) and separate Judicial Guidelines (for judicial officers), developed partly in response to a review by the Victorian Law Commission. The Guidelines confirm that generative AI must not be used for judicial decision-making, though Court-approved AI tools may be used to assist judicial officers and court staff, including to find case materials, produce summaries and chronologies from case materials, aid legal research, and proofread. The Guidelines stipulate that these permitted uses are not a substitute for reading or listening to evidence and submissions, or for fact-finding where required in judicial decision-making, and that judicial officers must consider each matter before them and exercise their own judgment in reaching decisions and giving reasons. |
Guidelines for legal practitioners by Australian courts
Courts of all Australian jurisdictions, apart from the Australian Capital Territory, the Northern Territory, and Tasmania, have published statements on the use of AI in legal practice.
|
Court |
Policy |
Content |
|
Federal Court of Australia |
Use of Generative Artificial Intelligence Practice Note (GPN-AI) (April 2026) |
The Federal Court recognises that generative AI has the potential to facilitate the resolution of disputes by increasing efficiency in the conduct of litigation, reducing legal costs, enhancing access to justice and the quality of the administration of justice, while stressing that it is fundamental to the administration of justice that generative AI must be used appropriately. The Practice Note flags particular risks around fictitious cases, citations or quotes, and misleading information on the law or how it might apply. On evidence, any use of generative AI must be consistent with the requirement that a person making an affidavit or witness statement represents that the document reflects their own recollection, knowledge, and/or experience, and an expert providing a report has an overriding duty to assist the Court impartially, with the report containing that expert’s own opinion and process of reasoning, subject to the disclosure obligations in the Expert Practice Note (GPN-EXPT (October 2016)). Disclosure of AI use is mandated in specific circumstances, and more broadly the Court retains a discretion to require disclosure wherever appropriate. |
|
Federal Circuit and Family Court of Australia |
Practice Direction: Use of Artificial Intelligence (PD-AI) (May 2026) |
Issued shortly after the Federal Court’s GPN-AI, the Federal Circuit and Family Court of Australia’s Practice Direction prohibits the input of confidential or privileged information into public AI tools and imposing strict requirements around discovery, disclosure orders, suppression orders, and subpoenas. |
|
Supreme Court of New South Wales |
Practice Note SC Gen 23 (January 2025) |
The Supreme Court of New South Wales has taken the strongest stance, introducing an absolute prohibition on the use of generative AI to draft affidavits, witness statements, character references, and any other material meant to reflect only the person’s own knowledge. With respect to other evidentiary material, such as expert opinions, the Supreme Court of New South Wales restricts the use of generative AI without prior leave. It allows such use only under conditions that ensure the expert’s reasoning process is clear to all parties. These conditions include disclosing that AI was used and keeping records of usage. |
|
Supreme, District and Planning and Environment Courts of Queensland |
Practice Direction 5 of 2025 (Supreme Court); 12 of 2025 (District Court); 7 of 2025 (Planning and Environment Court) - Accuracy of References in Submissions |
Rather than regulating AI use directly, Queensland’s approach fixes personal responsibility on a named practitioner. Issued by the Chief Justice of Queensland on 24 September 2025, the Practice Direction requires that the responsible person for submissions to the Court be identified and personally accountable for its content. By placing their name on a written submission, the legal practitioner informs the Court that they have verified the accuracy and relevance of any references to legislation, authorities or other sources, and ensured that the document is expressed in terms which reflect their judgment as to the proper discharge of their professional and ethical obligations, with the same obligation applying to oral submissions. Practitioners who submit references to non-existent cases or legislation may be referred to the Legal Services Commissioner for investigation and/or be required to show cause why a costs order should not be made against them personally. |
|
Supreme Court of South Australia |
Guidelines Concerning the Use of Generative AI in Litigation in South Australian Courts (January 2026) |
Following a profession-wide consultation, in January 2026 the Supreme Court of South Australia issued guidelines on generative AI use, which are expected to be given force under its Rules of Court. These apply not only to litigants but across the Supreme Court, District Court, Magistrates Court, Youth Court, Environment, Resources and Development Court, and the Court of Disputed Returns. The Guidelines encourage the appropriate use of generative AI to improve efficiency and access to justice, but make clear that existing duties of honesty, competence, confidentiality, and paramount duty to the court continue to apply in full, such that using generative AI is never itself a sufficient explanation or excuse for presenting inaccurate, misleading, or unsupported material to the court. |
|
Supreme Court of Victoria |
Practice Note SC Gen 25 - The Use of Artificial Intelligence by Court Users (May 2026) |
Developed following a report by the Victorian Law Reform Commission in October 2025, the Practice Note is framed around six principles: access to justice; impartiality and fairness; accountability and independence; privacy and data security; and transparency and open justice, among others. On evidence, the Practice Note requires particular caution: if generative AI tools are used to assist in the preparation of affidavit materials, witness statements or other documents which form the evidence or potential evidence of a witness, and that such documents be sworn or finalised in a form that reflects that person’s own knowledge and words. More generally, content produced using AI must be verified with meaningful human control, and court users are responsible for ensuring that content produced by AI is current, complete, accurate, and applicable to the jurisdiction. The Practice Note specifies that one AI tool cannot be used to verify the output of another. |
|
Supreme Court of Western Australia |
Guidelines for the Use of Generative Artificial Intelligence (November 2025) |
The Supreme Court of Western Australia’s Guidelines take an educative rather than prohibitive approach, cautioning users about generative AI’s risks (outdated training data, fabricated cases and authorities, jurisdictional mismatches and embedded bias) while making clear that any AI-generated content relied on in proceedings must be verified by a human who takes legal responsibility for it, such that existing professional and evidentiary obligations continue to apply. The Guidelines provide that, when directed by the court or where otherwise necessary or appropriate, the use of generative AI in proceedings (including the preparation of any materials) should be disclosed to other parties and the court. |
Guidelines for legal practitioners by law societies
The statements by the law societies or representative practitioner bodies of each Australian jurisdiction are broadly similar to those of Australian courts, revolving around a reminder of professional obligations; the limitations of AI in legal practice; and recommendations for what they view as best practice.
|
Body |
Policy |
Content |
|
Law Society of the Australian Capital Territory |
Use of AI in Legal Practice (October 2025) |
Based on a joint statement issued by the New South Wales Law Society, Victorian Legal Services Board and Commissioner and the Western Australia Legal Practice Board on 6 December 2024 (see below), the Australian Capital Territory guidance applies this to its own Legal Profession Act 2006 and Legal Profession (Solicitors) Conduct Rules 2015, with particular emphasis on maintaining client confidentiality under Rule 9 of the Conduct Rules, requiring that practitioners must not input confidential, sensitive, or privileged information into public generative AI tools. |
|
Law Society of New South Wales |
A Solicitor’s Guide to Responsible Use of Artificial Intelligence (November 2023, updated October 2024; January 2026) |
In November 2023, the Law Society of New South Wales issued a Guide to the Responsible Use of AI by Solicitors, cataloguing the professional duties most likely engaged by AI use, including competence, confidentiality, independence and candour to the court. The Guide cautions that solicitors should not rely on generative AI to verify sources produced by AI. |
|
Law Society of Western Australia |
The Law Society of Western Australia has prepared an AI Hub, aiming to support Western Australian legal practitioners by outlining the benefits, risks, and responsible use of AI. The Hub includes a page ‘Using AI: Your Questions Answered’, which answers a number of practitioner questions, reminding solicitors of their ethical obligations, and setting out the risks of using AI in practice. |
|
|
Law Institute of Victoria |
Ethics Guidelines - Ethical and Responsible Use of AI in Legal Practice (September 2025) |
The Law Institute of Victoria’s Guidelines provides an overview of best practices for solicitors using AI, including to never enter confidential client or legally privileged information, to disclose the use of AI to clients, and to check all AI-generated outputs for accuracy. The Guidance also provides that a practitioner is ultimately accountable for any documents tendered, and reminds solicitors of their ethical obligations. According to the Guidance, practitioners should exercise caution in using AI until they have received appropriate training. |
|
Queensland Law Society |
Guidance Statement No. 37 - Artificial Intelligence in Legal Practice (May 2024) |
The Queensland Law Society’s Guidance Statement frames principles rather than prescriptive rules on the responsible use of AI in legal practice, reminding solicitors to apply ethical duties when using AI and stressing that lawyers must exercise independent judgment, maintain confidentiality, ensure competence, and verify AI outputs. |
|
New South Wales Law Society, Victorian Legal Services Board and Commissioner and the Western Australia Legal Practice Board |
Joint Statement on the Use of Artificial Intelligence in Australian Legal Practice (December 2024) |
The Joint Statement on the Use of AI in Australian Legal Practice is the most significant cross-jurisdictional practitioner instrument, applying to solicitors in New South Wales and to solicitors and barristers in Victoria and Western Australia, and is grounded in the Uniform Law conduct rules common to those three jurisdictions. It provides, inter alia, that lawyers must maintain client confidentiality by not inputting confidential, sensitive or privileged information into public AI tools, providing independent advice with their own assessment and analysis, being honest and delivering legal services competently and diligently, and ensuring that costs remain fair, reasonable and proportionate. |
Guidance for legal practitioners by bar associations
In jurisdictions with a divided legal profession, state and territory bar associations have separately issued guidance to barristers on the use of AI, generally mirroring the framing adopted by Law Societies but drawing more directly on barristers’ specific duties of independence and candour, and non-deception to the Court under the applicable Barristers’ Conduct Rules.
|
Bar Association |
Policy |
Content |
|
Bar Association of Queensland |
Guidelines for the Use of AI in Legal Practice (November 2025) |
Explicitly tied to the Bar Association of Queensland’s Barristers’ Conduct Rules 2011 most engaged by AI use, the Guidelines for the Use of AI in Legal Practice provide, inter alia, that barristers must verify AI outputs, should be transparent about the use of AI, and should review the terms of service and user settings with a view to protecting their clients’ legal professional privilege and entitlement to confidentiality. |
|
New South Wales Bar Association |
Issues Arising from the Use of AI Language Models (including ChatGPT) in Legal Practice (July 2023) |
The New South Wales Bar Association Guidelines were the first bar-level AI guidance in Australia. They are framed around the New South Wales Legal Profession Uniform Conduct (Barristers) Rules 2015, and make clear that AI can never be used as a substitute for the proper exercise of a barrister’s professional judgment in matters of law or in ignorance of professional and ethical obligations, including the obligation to provide competent and diligent representation, to maintain independence and integrity, and to maintain the confidentiality of client information and privileged material. |
|
Victorian Bar Association |
Guidance on the Ethical Use of Generative AI (June 2026) |
As part of the Victorian Bar Association’s ‘Ethics Hub’ resources, which affirm that barristers must act in accordance with the Legal Profession Uniform Conduct (Barristers) Rules 2015, the Victorian Bar Association Ethics Committee published the Guidance on the Ethical Use of Generative AI to assist barristers with using AI tools responsibly. It emphasises the risks of using AI, particularly in relation to potential breaches of client confidentiality, and that barristers must disclose the use of AI to the court where required and be extremely cautious if generative AI is used to draft witness evidence or court submissions. |
Guidance for law enforcement
The Australian Federal Police has publicly confirmed in its AI Transparency Statement that they are committed to the Australia New Zealand Responsible and Ethical Artificial Intelligence Framework agreed upon by all Australian and New Zealand Police Commissioners. The Framework ‘serves as a foundational framework for the ethical, transparent, and effective use of AI in policing’ and aims to operationalise the ‘Australia New Zealand Artificial Intelligence Principles’ in the following ways:
|
Transparency |
Police should ensure clear and accessible information about the use of AI systems is publicly available. |
|
Human Oversight |
AI should only be used to inform, rather than replace, decision-making. Human oversight and control should be present ‘at all stages of the development, deployment and operation of the AI system’. |
|
Proportionality and Justifiability |
AI systems should be used by police in a reasonable, necessary, proportionate, and lawful manner that respects human rights. Any benefits to community safety outweigh prospective negative impacts resulting from the use of AI. |
|
Explainability |
AI systems should be appropriately described in a way that allows their use to be challenged. |
|
Fairness |
Police should design or use AI systems that respect equality, fairness, and human rights. AI systems should not be used to discriminate against individuals or groups, and biases should be screened for in risk assessments and subsequently managed. |
|
Reliability |
AI systems should be continuously monitored, tested, and developed to ensure optimal functionality and the attainment of intended purposes. |
|
Accountability |
Police organisations should possess layers of accountability and governance over the use of AI. |
|
Skills and Knowledge |
Members of police should have appropriate training and knowledge to competently deploy AI systems. This includes awareness of the limitations of AI and contemporary developments. |
|
Privacy and Security |
Privacy and security should be at the forefront of any use or design of AI systems. |
Separately, Victoria Police have published an AI Ethics Framework that is based on the application of eight enabling principles: human rights, community benefit, fairness, privacy and security, transparency, accountability, human oversight, and skills and knowledge.
Criminal procedure rules
The admissibility of AI-generated or AI-reviewed evidence in criminal proceedings is indirectly regulated by the Evidence Act of the relevant jurisdiction. In the Evidence Act 1995 (Cth), the Evidence Act 1977 (Qld), Evidence Act 1929 (SA), and the uniform Evidence Acts —that is, the substantially similar Evidence Acts of New South Wales, Victoria, Western Australia, Tasmania, the Northern Territory, and the Australian Capital Territory — the discretion generally ultimately rests on the court itself to draw their own inferences from documents regarding their authenticity and relevance as well as whether they ought to be excluded. Importantly, evidence can be excluded if it has a risk of misleading or confusing the jury to such an extent that it substantially outweighs its probative value and so serves as a filter for evidence regardless of whether AI was involved in its preparation.
Data protection legislation
The Privacy Act 1988 (Cth) implemented Australia’s obligations under the Organisation for Economic Cooperation and Development Guidelines on the Protection of Privacy and Transborder Flows of Personal Data and Article 17 of the International Covenant on Civil and Political Rights. Most Australian states and territories have privacy legislation which applies to their public sector agencies, while the Privacy Act 1988 (Cth) generally applies to Commonwealth agencies, individuals (including legal practitioners), and body corporates that have an annual turnover of more than $3,000,000. The Privacy Act 1988 (Cth) restricts the use of AI tools in criminal proceedings by prohibiting breaches of these entities that interfere with the privacy of an individual by breaching the Australian Privacy Principles set out by the Act in relation to personal information about the individual.
Thus, the Privacy Act 1988 (Cth, is a source of regulation for legal practitioners using AI, particularly given the vulnerability of public AI chatbots to store and recycle inputted material. Under the Act, an entity that is bound by the Australian Privacy Principles (‘APPs’)一which, depending on the practice’s structure, may include a law firm一commits an ‘interference with the privacy of an individual’ under Section 13 if it breaches an applicable APP. Several APPs may be directly engaged by the use of generative AI tools in legal practice:
- APP 6 restricts an entity to using or disclosing personal information only for the primary purpose for which it was collected;
- APP 10 requires reasonable steps to ensure the personal information an entity collects, uses or discloses is accurate; and
- APP 11 requires an entity to take reasonable steps to protect personal information it holds from misuse, interference, loss and unauthorised access or disclosure.
Individuals who believe that these Principles have been contravened may lodge a complaint with the Office of the Australian Information Commissioner. However, the Privacy Act 1988 (Cth) only applies to the administrative practices of the Federal Court, although it binds the Commonwealth Office of the Director of Public Prosecutions, legal practitioners, and law firms meeting the turnover threshold.
In a similar vein, there may also be issues with copyright infringement per the Copyright Act 1968 (Cth) if copyrighted material is used to train the AI tool or forms part of an AI database. At an individual level, clients may also have claims under the Australian Consumer Law in Schedule 2 of the Competition and Consumer Act 2010 (Cth) should large swathes of legal work be delegated to AI, such that the appearance of the provision of competent legal services may be misleading or deceptive.
Human rights
There is no federal Bill or Charter of Human Rights in Australia. However, three of the seven Australian states and territories have implemented human rights legislation, these are Victoria, the Australian Capital Territory and Queensland. In Victoria, Victorian courts (and the Victorian Civil and Administrative Tribunal) are obliged to give effect to human rights under the Charter of Human Rights and Responsibilities Act 2006 (Vic) in relation to their administrative functions rather than exercises of judicial decision-making. The Charter relevantly protects rights including recognition and equality before the law, the right to a fair hearing, the right to liberty and security of person, basic rights in criminal proceedings, and privacy and reputation. To the extent that the use of AI by Victorian courts infringes upon any of these rights, for example by making listing errors denying an individual of the right to a fair hearing, remedies are available if claimed as a cause of action and complaints can primarily be made to the Victorian Ombudsman.
Similarly, in the Australian Capital Territory, the Human Rights Act 2004 (ACT) applies to Australian Capital Territory Courts and the Queensland Human Rights Act 2019 (Qld) applies to Queensland courts when they are acting in an administrative capacity. It protects rights comparable to those in the Charter (above).
|
Right |
International |
Federal |
Domestic |
|
Privacy |
ICCPR Art 17 (ratified, not directly enforceable) |
Privacy Act 1988 (Cth) |
Privacy Act 1988 (Cth) — directly enforceable via APPs and OAIC complaints |
|
Fair Trial |
ICCPR Art 14 (ratified, not directly enforceable) |
No general statutory right |
Vic Charter s 24; Qld HRA s 31; ACT HRA s 21 (Vic/Qld/ACT only) |
|
Equality Before the Law |
ICCPR Art 26; ICERD Art 5(a) (both ratified, not directly enforceable) |
Racial Discrimination Act 1975 s 10 (directly enforceable); Sex/Disability/Age Discrimination Acts |
Vic Charter s 8; Qld HRA s 15; ACT HRA s 8 (Vic/Qld/ACT only) |
Outlook
The regulation of AI in criminal proceedings in Australia does not follow a unified national strategy. The existence of multiple independent jurisdictions has resulted in a fragmented regulatory landscape, with no overarching framework governing the use of AI across the criminal process. On 15 July 2026, the Prime Minister announced a new AI framework including immediate establishment of an Office of AI within the Department of Prime Minister and Cabinet and a commitment to develop Australian Standards for AI. It is intended to bring AI governance into a single, ‘clear, consistent and mandatory’ regulatory framework, however the focus initially does not seem to touch on judicial matters. The National Cabinet is expected to consider the approach in August 2026, with standards expected to be legislated in early 2027, which may impact the regulation of AI in criminal proceedings either directly or indirectly.
CASES
Australian courts have now encountered more than 30 cases in which legal practitioners' use of AI tools has resulted in false or fabricated information being filed or submitted to the court. In a number of these matters, judges have identified a strong public interest in referring the practitioner concerned to the relevant law society or legal profession regulatory body. The following quote by the Chief Justice of Western Australia demonstrates the initial reaction to the first case involving AI reported in Australia:
Mr Nash is unrepresented. He prepared the appellant's case himself, although it appears that he may have had some assistance with later submissions (including, perhaps, from an artificial intelligence program such as Chat GPT). Neither form of submission made coherent submissions as to why the trial judge's decision was affected by material error or otherwise gave rise to a miscarriage of justice. Nor did the material sought to be adduced by Mr Nash as additional evidence on the appeal disclose any miscarriage of justice. [. . .] Before briefly addressing the grounds of appeal, I first record that at the hearing of the appeal today, the Court dismissed the application dated 11 April 2023 to transfer the appeal to the Supreme Court of New South Wales (Court of Appeal). There was no merit in the application. First, there is no jurisdiction to transfer criminal proceedings. The Cross-vesting Act expressly excludes criminal matters. There is otherwise no jurisdictional basis to transfer criminal proceedings under State law in this Court to the court of another State. The authorities cited by Mr Nash in support of such jurisdiction do not exist; they are fictitious.
