In September 2024, we released the Voluntary AI Safety Standard (VAISS). This guidance, published in October 2025, is the first update of the VAISS.
In this update, we have:
- condensed 10 guardrails into 6 essential practices
- removed redundant language
- expanded our audience to developers as well as deployers.
The VAISS was the government’s first comprehensive AI governance resource to help organisations develop and deploy AI systems in Australia safely and reliably. It set out best‑practice AI governance, focused on AI deployers, that aligned with international standards and regulation.
The technology and governance landscape has shifted rapidly over the past year. The VAISS was intended to be iterative to ensure that this guidance remains fit‑for‑purpose. At time of release, we flagged that we would update the VAISS to extend best practices to AI developers. The National AI Centre started work on this next version of VAISS in late 2024.
NAIC received extensive feedback throughout 2024–25, as part of consultation extending VAISS practices to developers, including:
- Industry professionals more advanced in AI adoption, and technical experts from a wide range of organisations, valued the VAISS as a useful framework to compare and guide practices and procedures.
- Most industry stakeholders were seeking more accessible, actionable and streamlined guidance which can be tailored to both technical and non‑technical audiences, particularly SMEs.
- Specific additional guidance was requested on procuring AI systems as well as transparency mechanisms for AI‑generated content, such as watermarking.
In addition, the 2025 Responsible AI Index, released 26 August 2025, surveyed the state of responsible AI across a range of organisations and sectors. The report found that:
- Responsible AI practice adoption is progressing: 12% of organisations are now in the Leading category for implementing responsible AI practices, up 4% from 2024.
- A ‘saying‑doing’ gap remains: while 78% of respondents agreed with ethical AI performance statements, only 29% had implemented relevant responsible AI practices.
- Smaller organisations face challenges implementing more resource‑intensive governance practices: confidence levels in responsible AI declined for those organisations with 20–99 employees.
This guidance will underpin an expansion of NAIC’s tools and resources. We will roll this out the next 12 months. This integrated approach will build greater coherence and consistency in the delivery of advice to industry.
For organisations that used the VAISS, all practices have now been integrated into the Implementation practices. Please refer to the VAISS x Implementation practices crosswalk.