WA Health Artificial Intelligence Policy MP 0193/25 and Standard 6.1: Overall acceptable use: staff use only AI systems that have been assessed and approved, never in a way that could harm a person's fundamental rights, safety or autonomy; use and release WA health information in AI only as the Data Custodian authorises; never enter information or use AI output in a way that breaches privacy, copyright or intellectual property; safeguard personal information with secure techniques; disclose AI use where results could affect people, decisions or public opinion; avoid over-reliance and keep accountability for final decisions; scrutinise outputs for inaccuracy, bias and hallucination and report concerns; understand liability where use breaches contracts or terms; complete required training first; keep AI-created records under the State Records Act 2000; and understand the limits of generative AI using the government guidance

What does 6.1 require?

Staff may use AI systems only in accordance with the standard including its ethics principles, the policy frameworks and local policies and processes, and applicable law. In particular they use only AI systems that have been assessed and approved for use; never use AI in a way that could negatively affect a person's fundamental rights, safety or autonomy; use and release WA health system information in AI systems only as the Data Custodian has authorised; never enter information, or use AI-produced information, in a way that could breach privacy, copyright or intellectual property rights; where personal information is involved, protect privacy in line with legislation using secure data protection techniques; disclose that AI has been or will be used where the results could affect people, government or business decisions or public opinion; avoid over-reliance, remaining accountable and responsible for final decisions with AI supporting rather than replacing their work; always scrutinise outputs for inaccuracies, bias or hallucinations and report concerns; stay aware of liability where AI use contravenes contracts or terms of use or causes regulatory breaches, harm or inaccurate or inequitable outputs; complete any system-specific or mandatory training that is available before using a system; retain records the AI creates under the State Records Act 2000; and understand the limitations and risks of generative AI with reference to the state's interim guidance on public generative AI tools, generative AI for internal government use and large language models.