Australia - Australian Capital Territory - AI Policy (2025)

Overview

The Australian Capital Territory (ACT) Artificial Intelligence Policy serves as a foundational document for guiding the ethical, safe, and responsible integration and use of Artificial Intelligence (AI) technologies across the ACT Government. This policy is designed to ensure that the benefits of AI in enhancing public services, streamlining operations, and addressing complex challenges are realised, while simultaneously mitigating associated risks such as bias, privacy breaches, and security vulnerabilities. It represents a proactive stance by the ACT Government to establish clear guardrails for AI adoption, fostering public trust and aligning with broader national AI ethics principles. The policy acknowledges the rapid evolution of AI technology and positions itself as a dynamic framework that will continue to adapt to emerging risks and opportunities, ensuring that AI solutions are well-designed, secure, and appropriately governed within the territory.

Central to the ACT's approach is the recognition that AI, while offering transformative potential, must be deployed in a manner that respects human rights, upholds privacy, and ensures fairness and transparency. The policy is not intended to be the sole lens through which all AI projects are viewed but rather a crucial component of a comprehensive risk management strategy. It works in tandem with the ACT AI Assurance Framework, which provides practical tools and processes, including a self-assessment mechanism, to aid directorates in managing AI-related risks. By setting out clear expectations and responsibilities, the policy aims to build a culture of responsible innovation, where AI initiatives are rigorously assessed, continuously monitored, and subject to human oversight and accountability throughout their entire lifecycle.

Definitions

Within the context of the ACT Government Artificial Intelligence Policy, several key terms are defined to ensure a common understanding and consistent application of its provisions. Artificial Intelligence (AI) broadly refers to systems capable of performing tasks traditionally associated with human intelligence, such as learning, problem-solving, and decision-making. The policy's scope encompasses various AI technologies that the ACT Government may deploy to enhance public services and internal operations. A critical component is the ACT AI Assurance Framework, which is explicitly designed as a guiding structure for the responsible and safe deployment of AI. This framework incorporates a self-assessment tool, enabling government entities to systematically identify, evaluate, and mitigate risks associated with their AI systems, thereby ensuring adherence to the policy's objectives and fostering accountability.

Further essential definitions include the role of the AI Advisory Group (AIAG), an expert body established to offer guidance and oversight for AI initiatives across the ACT Government, working in conjunction with existing ICT governance structures. The policy also explicitly aligns with the National AI Ethics Principles, a set of overarching guidelines for ethical AI use in Australia, which underpin the ACT's commitment to developing trustworthy AI. Concepts such as 'Human Oversight' are paramount, emphasising the necessity for human review and intervention in AI-assisted decision-making processes. The policy addresses 'Bias in AI,' acknowledging the potential for algorithms to produce unfair or discriminatory outcomes and mandating strategies to minimise such risks. Lastly, 'Transparency' and 'Contestability' are defined as core tenets, ensuring that the public can understand AI's impact and challenge its outcomes, while 'Accountability' establishes clear responsibilities for those involved in the AI system lifecycle.

Governance and Institutional Framework

The governance and institutional framework underpinning the ACT Government Artificial Intelligence Policy is designed to ensure robust oversight and ethical deployment of AI technologies across all government operations. A cornerstone of this framework is the establishment of the ACT AI Advisory Group (AIAG). This group plays a crucial role in providing expert advice and strategic oversight for all AI initiatives within the ACT Government. The AIAG works collaboratively with existing ICT review and governance processes, ensuring that AI projects are not only technically sound but also align with the policy's ethical principles and risk management requirements. Its mandate includes reviewing AI initiatives, providing recommendations on project risk management, and granting approval to proceed through established project lifecycle governance.

Furthermore, the policy mandates that directorates within the ACT Government are responsible for managing AI risks through the consistent use of the ACT AI Assurance Framework's self-assessment tool. Completed self-assessments must be submitted to the AIAG for a contestability review, reinforcing a multi-layered approach to governance. This ensures that accountability is embedded at both the project level and through a central oversight body. The policy also highlights alignment with broader national standards, specifically Australia's AI Ethics Principles and the National AI Assurance Framework, ensuring that the ACT's approach to AI governance meets key national benchmarks and fosters community trust through well-designed, safe, and appropriately governed AI solutions. This integrated governance structure is crucial for navigating the complexities of AI adoption while maintaining public confidence.

Key Focus Areas

The ACT Government Artificial Intelligence Policy places significant emphasis on several key focus areas to ensure the responsible and ethical deployment of AI. A primary focus is on risk management, requiring directorates to identify, assess, and mitigate potential risks associated with AI technologies. This includes addressing issues such as algorithmic bias, privacy concerns, and security vulnerabilities that could undermine public trust or lead to adverse outcomes. The policy mandates the use of the ACT AI Assurance Framework's self-assessment tool as a core mechanism for managing these risks throughout the entire AI system lifecycle, from initial planning to ongoing operation. The objective is to proactively manage potential harms while harnessing AI's benefits for improved public service delivery.

Another critical area is transparency and accountability. The policy stipulates that there should be transparency and responsible disclosure, enabling individuals to understand when they are interacting with an AI system and how it might impact them. This commitment extends to ensuring that those responsible for different phases of an AI system's lifecycle are identifiable and accountable for its outcomes, with human oversight enabled to review and intervene in AI-driven decisions. Furthermore, data protection and privacy are paramount, with the policy emphasising that AI systems must respect and uphold individual privacy rights and ensure the security of data. This aligns with existing legislation such as the Information Privacy Act 2014 and the ACT Cyber Security Policy, which establish security standards for the ethical deployment of AI technologies and the protection of government data.

Implementation Framework

The implementation framework for the ACT Government Artificial Intelligence Policy is structured to ensure practical and consistent application across all ACT Government directorates. At its core, the policy establishes the ACT AI Assurance Framework as the primary mechanism for operationalising responsible AI use. This framework mandates that all AI initiatives falling within the policy's scope must be rigorously assessed at every stage of their lifecycle, from initial planning and development through to final delivery and ongoing operation. Directorates are required to utilise the assurance self-assessment tool provided within this framework to systematically identify, analyse, and manage potential risks associated with their AI systems. This structured approach ensures a consistent methodology for risk evaluation and mitigation across the government.

A crucial element of the implementation framework involves the ACT AI Advisory Group (AIAG). Following the completion of self-assessments, directorates must submit these to the AIAG for a contestability review. This review process provides an additional layer of scrutiny and expert guidance, with the AIAG either granting approval to proceed through regular project lifecycle governance or issuing recommendations for further project risk management. This collaborative approach ensures that AI initiatives benefit from collective expertise and adhere to the highest standards of ethical and responsible deployment. The policy also underscores the necessity for AI initiatives to comply with all relevant legislation, including the Information Privacy Act 2014 and the ACT Human Rights Act 2004, thereby integrating AI governance within the broader legal and regulatory landscape of the ACT.

Monitoring and Evaluation

The ACT Government Artificial Intelligence Policy incorporates robust mechanisms for the ongoing monitoring and evaluation of AI systems to ensure their continued safe, ethical, and responsible operation. The policy implicitly requires continuous assessment throughout the entire lifecycle of an AI system, acknowledging that risks and impacts can evolve over time. This includes regular reviews of services that utilise AI systems, ensuring that they consistently operate in accordance with their intended purpose and continue to meet community expectations. The comprehensive risk assessment associated with the ACT AI Assurance Framework is designed to evolve, capturing current and future risks as AI technologies advance and their applications become more complex. This adaptive approach to monitoring is crucial for maintaining the integrity and trustworthiness of AI deployments.

Furthermore, the policy's emphasis on accountability and transparency directly supports effective monitoring and evaluation. By requiring that individuals responsible for different phases of the AI system lifecycle are identifiable, it facilitates clear lines of responsibility for performance and adherence to policy. The mandate for transparency ensures that the significant impacts of AI are understood, which in turn enables informed evaluation by both internal stakeholders and the public. The ACT AI Advisory Group (AIAG) plays a key role in this, providing ongoing advice and oversight for AI initiatives. Through its contestability reviews of self-assessments, the AIAG contributes to the continuous improvement cycle, ensuring that any recommendations on project risk management are integrated and that AI systems are regularly scrutinised for their ethical implications, safety, and effectiveness.

Penalties, Liability, and Appeals

While the ACT Government Artificial Intelligence Policy primarily focuses on establishing a framework for ethical and responsible AI use, it implicitly addresses the implications of non-compliance through its emphasis on accountability and alignment with existing legal instruments. The policy states that AI initiatives must comply with all relevant legislation, which inherently means that any misuse or failure to adhere to established guidelines could lead to penalties or liabilities under existing ACT laws. For instance, breaches related to data privacy would fall under the purview of the Information Privacy Act 2014, which outlines provisions for handling personal information and potential consequences for non-compliance. Similarly, any AI system infringing upon human rights would be subject to scrutiny under the ACT Human Rights Act 2004, potentially leading to legal challenges and redress mechanisms.

The policy underscores the principle of accountability, stating that those responsible for different phases of the AI system lifecycle should be identifiable and accountable for the outcomes. This establishes a clear chain of responsibility, which is fundamental for determining liability in cases of AI-related harm or malfunction. Furthermore, the policy explicitly calls for a timely process to allow individuals to challenge the use or outcomes of an AI system when it significantly impacts them. This 'contestability' mechanism serves as a crucial avenue for appeals and redress, ensuring that individuals have a means to seek review or recourse against AI-driven decisions. While the policy itself does not enumerate specific penalties, its integration within the broader ACT legal and regulatory landscape means that existing laws and frameworks would apply to address issues of liability, non-compliance, and the provision of appeal mechanisms.

Relationship to Other Instruments

The ACT Government Artificial Intelligence Policy is designed to operate within a broader ecosystem of national and territorial legal and policy instruments, rather than as an isolated directive. A fundamental aspect of its design is its explicit alignment with Australia's AI Ethics Principles and the National AI Assurance Framework. This ensures that the ACT's approach to AI governance is consistent with national standards and contributes to a cohesive national strategy for safe and responsible AI. By adhering to these national frameworks, the ACT Government aims to build and maintain community trust in its AI solutions, ensuring they are well-designed, secure, and ethically governed in line with broader Australian expectations. This interconnectedness allows for shared learning and harmonised development of AI governance practices across different levels of government.

Beyond national AI-specific instruments, the policy also integrates with existing ACT Government legislation and policies. Notably, it requires AI initiatives to comply with the Information Privacy Act 2014, which governs the handling of personal information, and the ACT Human Rights Act 2004, which protects and promotes human rights within the territory. This means that AI system owners, particularly in sensitive domains like criminal justice or education, must consider potential infringements on individual rights. Furthermore, the policy references the ACT Cyber Security Policy, which establishes essential security standards for all ACT Government ICT systems, including those incorporating AI. This layered approach ensures that AI deployment is not only ethically sound but also legally compliant and cyber-secure, leveraging existing regulatory frameworks to provide comprehensive oversight.

International Alignment

The ACT Government Artificial Intelligence Policy, while specific to the Australian Capital Territory, demonstrates an inherent alignment with broader international principles and emerging best practices in AI governance. Although not explicitly detailing specific international treaties or frameworks, the policy's foundational commitment to ethical, safe, and responsible AI use, coupled with its alignment with Australia's National AI Ethics Principles, reflects a global consensus on the core values that should guide AI development and deployment. These national principles themselves often draw inspiration from international discussions and recommendations from bodies such as the OECD, which has been instrumental in shaping global dialogues on AI ethics and governance.

The policy's emphasis on human-centred values, fairness, privacy protection, reliability, transparency, contestability, and accountability echoes the key tenets found in many international AI ethics guidelines and regulatory proposals. For instance, the requirement for human oversight, the proactive management of algorithmic bias, and the provision for individuals to challenge AI-driven outcomes are consistent themes in global efforts to ensure trustworthy AI. By embedding these universally recognised ethical considerations, the ACT Government positions its AI policy in harmony with the evolving international landscape of AI regulation, demonstrating a commitment to responsible innovation that resonates beyond its jurisdictional borders and contributes to the global discourse on governing AI for the public good.

Implementation Timeline

Sources and References