What should I consider when implementing AI?

Safe AI implementation requires more than choosing the right technology.

Governance and accountability
Make sure responsibilities, decision-making and oversight arrangements are clear throughout implementation.
Data
Consider the quality, suitability, security, privacy and governance of the data your AI system relies on.
Security
Identify and manage security risks associated with AI systems, models, infrastructure and data.
Testing and evaluation
Test AI systems before deployment and establish appropriate measures for performance, safety and reliability.
Deployment and integration
Consider how AI will integrate with your existing systems, processes and operating environment.
Monitoring and review
Plan for ongoing monitoring so your agency can identify changes in performance, emerging risks and unintended impacts.

Building AI responsibly across the lifecycle

Put AI into practice safely, securely and responsibly using the AI policy, AI impact assessment tool, technical standards, guidance and procurement resources available to Australian Government agencies.

Whether you're developing, procuring or deploying an AI system, consider its impacts and risks, along with technical, security, data, governance and operational requirements throughout implementation, not just before deployment.

Guidance for policy, governance and risk

Assessing AI risk?
The AI Impact Assessment Tool helps agencies meet this requirement. It supports them to identify, assess and manage impacts and risks against Australia’s AI Ethics Principles and across broader governance, technical and operational considerations.
AI Review Committee
The AI Review Committee provides expert and non-binding advice on AI use cases with a residual high-risk rating and those that agencies determine to be highly sensitive, novel or complex. Understand more about the Committee and the process for seeking review.
Implementing the AI Policy?
The AI Policy provides the foundation for safe AI implementation. In-scope use cases require an impact assessment and should align with policy requirements and agency governance arrangements.

Technical guidance

Using the AI technical standard?
The AI technical standard provides a consistent approach to designing, developing, deploying and operating government AI systems. Use the standard to address key technical considerations across the AI lifecycle.
Using public generative AI?
Publicly available generative AI tools can introduce particular security, privacy, and data considerations. Agency guidance is available to help Australian Government agencies manage the use of public generative AI tools safely and responsibly.
Using agentic AI?
Agentic AI systems can introduce additional risks because they may operate with greater autonomy and act on behalf of users. Use the agentic AI addendum with the technical standard to understand these additional considerations.

Procuring AI?

When buying AI products or services, consider technical, security, data, governance and operational requirements. Use BuyICT to access Australian Government procurement information and marketplaces for digital and ICT products and services.
 

Where do I go next?

Once your AI solution is being implemented, focus on building workforce capability so staff have the skills, knowledge and confidence to use, manage and oversee AI effectively throughout its lifecycle.

Understand the AI policy
Familiarise yourself with the AI Policy and supporting guidance and resources.
Assess AI use cases
Identify opportunities and assess their value, impact, risk and suitability.
Meet AI technical requirements
Use the AI technical standard to help ensure your AI systems are secure, reliable and fit for purpose.
Build capability
Make sure the people developing, managing and using AI have the skills and knowledge they need.

Connect with the digital community

Share, build or learn digital experience and skills with training and events, and collaborate with peers across government.

Join the Digital Profession