AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 7

Office of the Australian Information Commissioner (OAIC)

Submitter group: Regulators.

An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.

Open in the interactive guide

Summary

Read the original submission for the full text.

Describes how the Privacy Act and FOI Act apply to AI, the OAIC's regulatory activities on AI training, AI scribes and government automated decision-making, and community attitudes survey results. Argues that AI's productivity benefits depend on guardrails and public trust. Supports Privacy Act reforms, a consistent government ADM framework, and rights preservation in APS AI use.

Key points

Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.

  1. The OAIC argues that community trust built through transparency is a precondition to benefiting from AI, and that privacy and information access frameworks provide the scaffolding for adopting it.

    The OAIC believes community trust built through transparency and the promotion of information rights is a precondition to fully harnessing the benefits and opportunities of AI, and that the privacy and information access regulatory frameworks provide the scaffolding within which entities can confidently adopt and innovate with AI.

  2. States that privacy obligations cover personal information used in AI training, testing, development and deployment, and information-access obligations apply when AI informs government decisions.

    Privacy obligations apply where personal information is handled to train, test, develop and deploy AI systems. Information access obligations apply where AI is used to inform government decision-making and involves mandated proactive release, informal release, and reactive release in response to a formal FOI application.

  3. Argues that government agencies need good records of AI inputs and outputs to support explainability of decisions involving AI.

    This particularly raises the need among government agencies for good record-keeping practices and hygiene around the inputs and outputs of AI systems to support the explainability of decisions involving AI

  4. Says 80% want to be informed when AI is used and 74% want notification if their personal information will be shared with third-party AI providers.

    80% want to be informed when AI is being used and 74% want to be told if their personal information will be shared with third-party AI providers

  5. Says Tranche 2 Privacy Act reforms aim to balance stronger privacy protections with innovation, including addressing emerging AI risks.

    The OAIC welcomes the announcement of Tranche 2 reforms to the Privacy Act. 19 The proposed measures aim to strike the right balance between strengthening privacy protections and supporting innovation in a contemporary digital context, including by tackling emerging AI risks to ensure the productivity benefits can be fully realised.

  6. Says it supports a consistent legal framework governing automation in government services.

    The OAIC supports the development of a consistent legal framework to govern the use of automation in government services.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Incorporate anticipatory regulation into systems supporting innovation and rights preservation: guiding values, strategic intelligence, stakeholder engagement, agile regulation and international cooperation.

    The elements of anticipatory regulation should be incorporated in any system designed to support innovation and assure rights preservation. These elements include guiding values, strategic intelligence, stakeholder engagement, agile regulation and international cooperation.

  2. Support responsive regulation through domestic information sharing and central notification systems.

    Domestically sharing information and central notification systems will be key to promoting a responsive regulatory environment.

  3. Strengthen the Privacy Act for new and emerging technologies.

    The Australian Government has announced plans for further strengthening the Privacy Act to deal with new and emerging technologies, which the OAIC strongly supports (see below).

  4. Publish IPS information on ADM-authorising legislation, use of ADM for public information and services, decisions with understandable examples, and policies governing decisions and recommendations affecting the public.

    For agencies using ADM, better practice to promote transparency and compliance with the FOI Act is to ensure IPS information covers: • The legislation that grants agencies the power to use ADM, and whether or not they utilise ADM to provide information and services to the public • A list of decisions agencies use ADM for and relevant and easy to understand examples, so the public can better understand how it is being utilised to make decisions that affect them • Policies that set out the principles for when and how agencies use ADM to make decisions and recommendations affecting members of the public.

  5. Develop a consistent legal framework governing automation in government services.

    The OAIC supports the development of a consistent legal framework to govern the use of automation in government services.

  6. Require agencies to record ADM inputs, their provenance and accuracy, impacts of outputs on decisions, and procedures ensuring output accuracy.

    Obligations for agencies to record information inputted into an ADM system, the provenance and accuracy of inputs, the impacts of ADM outputs on decisions, and procedures in place to ensure the accuracy of ADM outputs

  7. Require proactive disclosure of when and how ADM is used and the assurance processes ensuring lawful use.

    Proactive disclosure around when and how ADM is used, and the assurance processes in place to ensure ADM is being used lawfully

  8. Create clear agency duties to explain ADM systems meaningfully on request, enabling individuals to exercise review rights.

    Clear duties for agencies to, upon request, provide meaningful explanations of ADM systems to enable individuals to exercise their review rights

  9. Establish a register of AI and ADM system use by Commonwealth agencies.

    Establishing a register for the use of AI and ADM systems by Commonwealth agencies

  10. Reform agencies' obligations and responsibilities when using contracted service providers for outsourced ADM systems, ensuring information rights are upheld.

    Noting that many ADM systems are outsourced, we stressed the need for reforms that address the obligations and responsibilities of agencies when using contracted service providers to ensure that information rights are upheld.

  11. As the whole-of-government ADM framework commences, ensure agencies meet its obligations alongside existing IPS publication obligations.

    We emphasise that IPS obligations for agencies to proactively publish operational information and decision-making in respect of ADM are extant and that they should work towards implementing the recommended practices for promoting transparency outlined in our report on Automated decision- making and public reporting under the Freedom of Information Act apply. 22 As the whole-of-government ADM framework commences, agencies should ensure they meet the obligations of both regimes.

  12. Agencies should promote information rights proactively throughout AI procurement, development and deployment rather than treating preservation as a one-off exercise.

    The statement is a call to action for agencies around the strategic need for a proactive approach to promoting information rights in the procurement, development and deployment of AI – rather than a one-off ‘set and forget’ approach.

Read the original

Open in the interactive guide