8 AI Ethics Principles

What it is

Australia's AI Ethics Principles are eight voluntary, aspirational principles for the responsible design, development, deployment and use of AI. They were published by the Department of Industry, Science and Resources (DISR) in November 2019 following public consultation, and they remain the conceptual baseline for Australian AI governance discourse. The CSIRO AI6, the NSW AI Assessment Framework, the Voluntary AI Safety Standard and the (now-shelved) mandatory guardrails all reference back to the eight principles in one form or another.

The eight principles

• Human, societal and environmental wellbeing. AI systems should benefit individuals, society and the environment.
• Human-centred values. AI systems should respect human rights, diversity, and the autonomy of individuals.
• Fairness. AI systems should be inclusive and accessible, and should not involve or result in unfair discrimination against individuals, communities or groups.
• Privacy protection and security. AI systems should respect and uphold privacy rights and data protection, and ensure the security of data.
• Reliability and safety. AI systems should reliably operate in accordance with their intended purpose throughout their lifecycle.
• Transparency and explainability. There should be transparency and responsible disclosure so people can understand when they are being significantly impacted by an AI system, and can find out what the AI system is doing.
• Contestability. When an AI system significantly impacts a person, community, group or environment, there should be a timely process to allow people to challenge the use or outcomes of the AI system.
• Accountability. Those responsible for the different phases of the AI system lifecycle should be identifiable and accountable for the outcomes of the AI systems, and human oversight of AI systems should be enabled.

Voluntary, but durable

The principles have no statutory force. No agency enforces them, no penalty attaches to ignoring them, and adoption is entirely at the organisation's discretion. They have nonetheless proven durable as the reference point Australian regulators, procurement officers and standards bodies use when describing what good looks like. The 2024 Responsible AI Index found a roughly 49-point difference between the proportion of organisations that said they were following the principles and the proportion that could demonstrate the practices behind them, which is part of what motivated the eventual move toward more structured guidance like the AI6 and the Voluntary AI Safety Standard.

How they relate to other frameworks

The eight principles sit at a high level of abstraction. They tell you what AI governance is supposed to achieve without telling you how to achieve it. The CSIRO AI6 picks up where the principles leave off by translating them into six operational practices, the NSW AI Assessment Framework operationalises them into a graded risk classification system, and the Privacy Act amendments make a specific subset of the transparency principle legally enforceable for ADM. So the eight principles are best understood as a vocabulary, with the other frameworks supplying the grammar.

Common misreadings

"They're just words, not a real framework." They are aspirational and voluntary, but they're also the lexicon every other Australian AI governance instrument is built on. An organisation that hasn't internalised the eight principles will struggle to read the more specific frameworks, because those frameworks assume the vocabulary.

"They overlap, so we only need to address some of them." The principles overlap deliberately, because real-world AI governance failures usually touch several at once. A failure of fairness almost always involves transparency and contestability problems too. Treating them as a checklist where partial coverage is enough misses the point.

"We have a privacy policy, so privacy protection is covered." The privacy protection principle is broader than the Privacy Act. It covers data protection through the AI lifecycle, not just disclosure to data subjects. A compliant privacy policy is a starting point, not a finishing line.

How Aicura supports work against it

Aicura uses Australia's AI Ethics Principles as a guidance source in two places. First, the Governance Policies scanner uses the principles when reviewing AI Ethics and Human Rights Policies, since that policy type is the most direct organisational response to the eight principles. Second, the principles sit inside the curated risk content that drives per-system Risk Assessments, providing the high-level vocabulary that anchors the more specific framework content. The Template Library's AI Ethics and Human Rights Policy generator produces drafts structured around the eight principles.

A note on this page

This is Aicura's reading of the framework, written to help you understand what the product is wired into. It is not legal advice and it is not the framework itself. Read the primary source. Where Aicura's interpretation differs from yours or from your advisors', go with theirs.

For a more complete picture of what Aicura does and doesn't do for any framework, see the boundaries page.