Transparency of Automated Decision Making (AU0024)

Summary

The purpose of this commitment is to ensure greater transparency about the use of automated decision-making (ADM) in government and contribute to safe and responsible Artificial Intelligence (AI) while facilitating innovation. This is intended to be achieved through: • improved governance, capability and guidance to support the safe and responsible use of AI in Australia • consideration of a consistent legal framework governing automated decision making by Australian Government agencies, and • enhanced transparency and protections for personal information handling by entities subject to the Privacy Act 1988.

Why this commitment is important

Delivery of this commitment aims to address community concerns about the transparency and integrity of decisions made using ADM systems, and about the safe and responsible use of AI. It is important that there is a consistent legal framework to support the use of automated decision-making, so that government services can operate ethically, without bias and with appropriate safeguards. There is also a need to support the transparency and integrity of decisions made using ADM systems including that individuals should be able to request meaningful information about how automated decisions, with a legal or similarly significant effect, are made when using personal information. These objectives are reflected both in the Government’s response to the Robodebt Royal Commission which committed to consider options for a consistent legal framework for ADM, and the Government’s response to the Privacy Act Review Report. Greater transparency about the use of ADM is part of fostering the public trust Australia needs to fully realise the benefits of AI, including opportunities for enhancing productivity, facilitating economic growth, and improving outcomes for Australians across health, education, environment and national security. This work will support the Australian Government to make advancements in this rapidly developing area and establish governance mechanisms to ensure AI is developed and used safely and responsibly in Australia. These mechanisms can include regulations, standards, tools, frameworks, principles and business practices.

Related OGP challenge areas

• Digital governance • Access to Information

How the impact of this commitment can be measured

• The Australian Government is able to demonstrate transparency in the use of ADM. • Governance mechanisms are established to ensure AI is developed and used safely and responsibly in Australia.

Other relevant information e.g. inquiries, research work being undertaken

The Department of Industry, Science and Resources (DISR) recently held public consultation on Safe and Responsible AI, which resulted in more than 500 responses reaffirming community expectations around transparency in the use of AI technologies, particularly by government. The Australian Government is currently reviewing these submissions. On 19 September 2023, the Hon Ed Husic MP, Minister for Industry and Science, and Senator Katy Gallagher, Minister for Finance, announced the establishment of the AI in Government Taskforce to focus on the safe and responsible use of AI within government. The Taskforce reports to the Secretaries Board Future of Work Sub-committee and is jointly led by the Digital Transformation Agency (DTA) and DISR. It is developing approaches for governance, risk management, skills and capability, and technical use and preparedness for AI at a whole-of-government level. The Attorney-General’s Department is progressing work on the Government Response to the Privacy Act Review Report, which was released on 28 September 2023, which includes agreement to proposals regarding the use of ADM (proposals 19.1–19.3). The Government Response to the Robodebt Royal Commission committed to considering opportunities for legislative reform to introduce a consistent legal framework in which automation in government services can operate ethically, without bias and with appropriate safeguards, which will include consideration of review pathways and transparency mechanisms (recommendation 17.1). OGP Global has further recommendations on digital governance and automated decisionmaking in their Open Government Guide.

Milestones*

1.1 Taskforce Update AI in Government Taskforce to update the Interim Guidance on the use of Generative AI for Agencies.

1.2 Government Response Release Government Response to the Privacy Act Review Report

1.3 Government Response Release Government Response to the Robodebt Royal Commission

1.4 New legislation Introduce legislation to protect the personal information of Australians, including in relation to the use of ADM.

1.5 Draft policy development AGD to lead policy development and scoping process for options to develop a comprehensive and consistent whole-of-government legislative framework for ADM in response to Robodebt Royal Commission recommendation 17.1, including consultation within government and targeted consultation with external stakeholders. This will include consideration of options for automation in government services to operate ethically, without bias and with appropriate safeguards.

1.6 Public consultation AGD to lead public consultation and advice on options to implement a whole-of-government legislative framework for ADM, and finalise legislation for Government consideration and introduction.

1.7 Implement new framework AGD to lead implementation of any whole-ofgovernment legislative framework for ADM, including developing guidance material and providing advice to agencies.

*Additional or more detailed milestones may be added in 2024, following Government consideration.

Action Plan Review

Commitment 1. Automated Decision-Making Transparency

Commitment 1. Automated Decision-making Transparency. Attorney-General’s Department; Office of the Commonwealth Ombudsman.

For a complete description of the commitment, see commitment 1 in Australia’s Open Government Partnership National Action Plan 2023-2025.

Context and objectives:

This commitment, proposed by government members, but reflecting a growing public concern acknowledged by OGP civil society members, [1] aims to address wide concerns across Australia about decisions the government makes using automated decision-making (ADM). Aligning with government work already underway, it responds to Recommendation 17.1 in the July 2023 Royal Commission report on the Robodebt scheme. [2] Over a period of six years, the scheme automatically matched data welfare recipients provided to Centrelink with data from the Australian Tax Office and sent out letters erroneously demanding people pay back thousands of dollars to the government. This resulted in very serious social consequences, including cases of suicide.

The commitment’s first three milestones were implemented in 2023, prior to this action plan’s formal implementation period: releasing the Artificial Intelligence (AI) in the Government Taskforce’s updated Interim Guidance on the use of public generative AI tools for agencies in November, [3] releasing the government’s response to the Privacy Act Review report in September, [4] and releasing the government’s response to the Royal Commission’s Robodebt report in November. [5] Both responses plan for legal reforms that the commitment intends to carry out during the implementation period.

Potential for results: Substantial

This commitment seeks to re-establish public trust in government’s use of ADM through new legislation protecting the use of personal information in ADM decisions, as well as a legislative framework for ADM. This is a priority for Australians, as evidenced by the volume and range of submissions to the Royal Commission and the issues raised in the Senate’s earlier public inquiries on the Robodebt scheme. [6] According to the Law Council, the community sector, including community legal assistance, and the many Australians who were affected by Robodebt, were responsible for bringing the issues raised by the scheme to the fore. [7] When accepting the recommendations of the Royal Commission, the Government Services Minister emphasized the importance of ensuring Robodebt cannot happen again, recognizing the toll it took on those who took their own lives, and on their families. [8] The legislation to be introduced by this commitment is expected to ensure that individuals can request meaningful information about how automated decisions are made when using personal information.

Initially, the commitment plans to introduce new legislation protecting personal information, including when ADM is used, in 2024. The Attorney-General’s Department (AGD) has clarified that the legislation will be accompanied by guidance to agencies and that the Office of the Information Commissioner (OAIC) will continue to advise the public on this matter. [9] The Privacy Act Review response committed the Government to reforms that will provide stronger privacy protections while encouraging digital innovation. [10] Proposal 19.1 of the government’s response requires regulated entities to ensure that their privacy policies set out the types of personal information that will be used in “substantially automated decisions” which have a legal or otherwise significant effect on an individual, and Proposal 19.3 proposes that individuals have a right to request meaningful, jargon-free, and clear information about how automated decisions are made that have a legal (or similarly significant) effect on an individual’s rights. [11] The media notes that these proposals are modelled on Article 22 of the European Union’s General Data Protection Regulation (GDPR) [12] but would apply to a wider range of automated decision-making than under the GDPR, which applies to solely automated, rather than substantially automated, decisions. [13]

In 2025, the commitment plans to introduce and implement a transparent whole-of-government legislative framework for ADM, focused on ethical and safe automation in government services, along with developing guidance material and advice to agencies. Led by AGD, it will be developed through consultation with government, targeted external stakeholders and through public consultation in 2024-25. The government’s response to the Royal Commission stated that it will consider opportunities to ensure a consistent clear and transparent legal framework for making government decisions involving automation, harnessing technological advances in artificial intelligence and automation, and develop a comprehensive and consistent legal framework to support automated decision-making, consistent with the principles recommended by the Royal Commission. [14]

Civil society members of OGF who provided feedback on this commitment recognized the scale and clarity of its milestones, through the proposed legislation introducing binding change. [15] Although the commitment does not offer assurances on the particular scope of intended legislation, in May 2024, the Attorney General described the scope of the Privacy Act reform which will be introduced in August 2024. [16] The government has agreed that the types of personal information to be used in substantially automated decisions which have a legal or similarly significant effect on an individual's rights should be clearly outlined in privacy policies. [17] In June, the Administrative Review Tribunal Act 2024 was enacted. This Act replaced the Administrative Appeals Tribunal with the Administrative Review Tribunal and re-established the Administrative Review Council to support ongoing improvements to administrative decision-making across government. [18] As an independent body, the Council has the power to determine matters for inquiry relating to the Council’s functions. The May 2024 announcement of the next Australian Information Commissioner continues to provide assurance that government guidance ensuring appropriate promotion and protection of information access and privacy rights will continue to be a top priority. [19]

Should the government continue to progress this level of legislative change relating to governing ADM in government, enhancing transparency and protecting personal information, and improving non-legislative governance capability and guidance, the potential for results from this commitment would be substantial.

Opportunities, challenges, and recommendations during implementation

The government is well placed to progress this commitment’s ADM transparency. Under Australia’s Drafting Directions, the Transparency and Administrative Law (TAL) Branch of the AGD is empowered to design the commitment’s new legislation. [20] The commitment will also benefit from parallel work by the AI in Government Taskforce. [21] In addition, the government established a new Artificial Intelligence Expert Group in February 2024 to provide advice to the Department of Industry, Science and Resources on immediate work on transparency, testing, and accountability, including options for AI guardrails in high-risk settings. [22] The main challenge for this commitment is to formulate legislation that addresses both privacy protection and the open government value of transparency.

Extra non-legislative milestones, which build on the government’s parallel ADM work, could assist with developing and implementing a whole-of-government ADM policy framework, regardless of whether the ADM legislative framework is passed in full. To support this commitment’s implementation and to fully utilise this action plan’s two-year period, the IRM recommends: