NSW has introduced new guidelines to support the safe and ethical use of agentic artificial intelligence (AI) in government work. The guidelines aim to help agencies use AI to solve complex problems and improve services for the community.
Agentic AI can make decisions and plan actions, unlike automation, which follows pre-set rules. The technology is already being used in some banks to detect fraud and in online stores to help merchants manage products and digital layouts.
The guidelines provide clear steps for responsible AI use. Agencies must define when human oversight is needed, review AI outputs regularly, and apply privacy and security measures. Checklists help determine if an AI agent is required and guide preparation for pilot projects and full deployment.
Programs using agentic AI will be assessed using the AI Assessment Framework. Projects with higher risks will need review by the AI Review Committee. The framework is being updated to include specific guidance for AI agents.
The Office for AI will continue to refine guidance as global AI standards evolve. The aim is to maintain human control while using AI safely and effectively.
The agentic AI guide is available online at: www.digital.nsw.gov.au/policy/artificial-intelligence/guide-to-using-ai-agents-nsw-government