The National Artificial Intelligence Centre of the Australian Government and research firm Fifth Quadrant unveiled the Responsible AI Index 2025 in Canberra. This index evaluates how Australian organisations incorporate artificial intelligence into their operations. It focuses on key principles like accountability, safety, fairness, transparency and explainability across four distinct maturity levels. Introducing Advancing Execution and Guiding Findings shows that 12% of organisations classify themselves as leaders, reflecting a four-point increase from 2024. 

Meanwhile, 48% are in the Developing stage and 17 percent are categorised as Emerging. The average RAI score is 43 out of a possible 100. The Centre stated that organisations leading in AI reap the rewards of responsible AI. The Index shows how responsible AI enhances digital government, improves customer experience, boosts employee engagement, increases productivity, and strengthens risk resilience.

Index evaluates adoption

Researchers evaluated 418 Australian organisations to examine how they integrated responsible artificial intelligence practices into their operations for the year 2025. The assessment evaluated adoption based on five key dimensions: accountability and oversight, safety and resilience, fairness, transparency and explainability, and contestability. We assigned equal weight to the evaluation criteria, allowing for a maximum of 20 points per dimension and resulting in a total possible index score of 100. 

Participants underwent assessment based on 45 practices that correspond to the Voluntary AI Safety Standard introduced in 2024. The standard establishes ten key guidelines that encompass risk management, data governance, transparency and human oversight. The national mean score for 2025 is 43, showing a decline of one point from 2024 and indicating inconsistent progress. The analysis shows that 48 percent of organisations fall into the Emerging category, 23 percent into Developing, 17 percent into Implementing, and 12 percent into Leading. 

Prominent organisations adopt more than 30 responsible AI practices and share insights from more than four years of AI utilisation across six technologies. Emerging organisations generally implement fewer than five practices and face challenges with oversight and limited experience in AI. The findings indicate a need for agencies to improve data governance, invest in secure data infrastructure, and address issues related to privacy, sharing, and cybersecurity. Not developing capabilities for Cloud Storage and data collaboration may hinder public sector institutions in their responsible adoption efforts and restrict their capacity to provide reliable services to citizens.

Index reveals gaps

• The mean Responsible AI score for 2025 is 43 out of 100, showing a decline of one point from 2024. This result shows a gradual overall improvement and highlights that many agencies still lack strong governance frameworks. AI systems in the public sector raise significant concerns about service delivery risks, especially without well-defined accountability frameworks.
• The maturity distribution shows that 48% of organisations fall into the emerging category, 23% are classified as developing, 17% are in the implementing stage, and 12% are recognised as leading. Major organisations led the adoption process with their robust data infrastructures and governance capabilities. Smaller agencies often face challenges from constrained budgets and reliance on outdated systems. This disparity illustrates the potential dangers of inconsistent service standards across different jurisdictions.
• Companies that have used AI for more than four years have seen significant benefits. Sixty percent report better customer experience, fifty-six percent note increased employee engagement, and forty-seven percent experience productivity improvements. This leads to quicker processing times for citizen services, improved staff distribution in frontline positions, and increased responsiveness in times of crisis.
• Documentation practices gain more prevalence than safety or supply chain evaluations. The current disparity risks privacy violations for government agencies, creates cybersecurity challenges, and exposes them to potential dangers from unverified third-party vendors. Effective measures require enforcing supply chain audits and adopting cybersecurity frameworks designed to tackle risks linked to AI.

Secures public sector AI

The ResponsibleAI Index 2025 shows that many organisations focus on transparency initiatives, such as documentation and disclosure, but fall short of establishing governance oversight and conducting supply chain assessments. Sixty-three percent of organisations indicate they use transparent practices, but fewer than 30 percent implement robust governance frameworks or independent review mechanisms. Human oversight remains insufficient. Less than one in three organisations implements practices that enable employees to intervene when AI decisions pose substantial risks. 

The public sector faces significant challenges because of this imbalance, which leads to vulnerabilities in data governance, privacy, and security. Organisations that gather and disseminate large amounts of citizens’ data may reveal sensitive information if they implement AI systems without strong privacy safeguards and tailored cybersecurity protocols. Data silos worsen the issue by hindering collaboration among agencies and restricting access to dependable data sets across various departments. 

This undermines the ability to create AI systems that are equitable and responsible and transparent. Smaller agencies fall behind in adopting cloud storage and modern data infrastructure. Their ability to establish secure data-sharing frameworks is limited, which is essential for digital government initiatives like comprehensive service portals or inter-agency emergency response systems. Not addressing these gaps may cause government organisations to undermine public trust and miss the opportunity to leverage artificial intelligence to enhance service delivery, resilience and transparency.

Index guides leaders

The Responsible AI Index 2025 shows that many organisations focus on transparency initiatives, such as documentation and disclosure, but fall short of establishing governance oversight and conducting supply chain assessments. Sixty-three percent of organisations indicate they use transparent practices, but fewer than 30 percent implement robust governance frameworks or independent review mechanisms. Human oversight remains insufficient. Less than one in three organisations implements practices that enable employees to intervene when AI decisions pose substantial risks. 

The public sector faces significant challenges because of this imbalance, which leads to vulnerabilities in data governance, privacy, and security. Organisations that gather and disseminate large amounts of citizen data may reveal sensitive information if they implement AI systems without strong privacy safeguards and tailored cybersecurity protocols. Data silos worsen the issue by hindering collaboration among agencies and restricting access to dependable data sets across various departments. 

This situation undermines the ability to create AI systems that are equitable and responsible and transparent. Smaller agencies fall behind in adopting cloud storage and modern data infrastructure. Their ability to establish secure data-sharing frameworks is limited, which is essential for digital government initiatives like comprehensive service portals or inter-agency emergency response systems. Not addressing these gaps may cause government organisations to undermine public trust and miss the opportunity to leverage artificial intelligence to enhance service delivery, resilience and transparency.

The Responsible AI Index 2025 shows that Australian organisations are making progress, but a national mean score of 43 reveals inconsistent adoption, and many agencies still struggle to manage risks effectively. Public sector leaders must enhance data governance frameworks, invest in robust data infrastructure, and ensure secure cloud storage. 

It is essential to minimise dependence on data silos to fully realise the advantages of artificial intelligence. Organisations that adhere to the Voluntary AI Safety Standard and incorporate oversight, safety, and fairness into their systems will deliver reliable services. The vision is clear: responsible AI serves as a necessity and the cornerstone of digital governance. Developing secure, transparent and efficient services that can evolve to meet future challenges is crucial.

Editors Publicspectrum

Website |  + postsBio

Public Spectrum is the first knowledge-sharing platform in Australia to embrace the entire public sector. This website is a platform where you can connect, collaborate, empower, inspire, and upskill with public sector professionals.

• Editors Publicspectrum

  Petra Andrén on how Horizon Europe can help Australia scale deep tech
• Editors Publicspectrum

  Privacy Commissioner rules tracking pixels breached patient privacy 
• Editors Publicspectrum

  Out-of-home advertising grows as brands rethink digital strategy 
• Editors Publicspectrum

  $30 million investment targets regional mobile blackspots across Australia