WA health entities must employ only AI models that can be explained and understood by developers, health care professionals, patients, users and regulators, avoiding tools where the type of tool and how it was trained and proven to work have not been disclosed. They must always be open and transparent when using AI systems that could significantly affect a person, community, group or the environment, conveying material risks and benefits so patients and others can make informed decisions and giving end users, patients and responsible parties enough to understand how the algorithm makes decisions. They seek a patient's consent to AI use unless an exception under MP 0175/22 Consent to Treatment Policy applies, and patients should be able to opt out. To build trust they disclose, at a minimum, when generative AI tools are used to inform government activities or decisions, to generate official briefings or corporate communications, or to enable autonomous decision-making, with disclosure statements covering how the AI contributed (prompts entered and resources used), the human oversight or editing applied afterwards, and the version and type of model. Where language or cultural barriers may impede transparency, staff refer to MP 0051/17 Language Services Policy and work in partnership with culturally and linguistically diverse and Aboriginal communities to evaluate and document the appropriate level of communication. They comply with Information Privacy Principle 10 of the Privacy and Responsible Information Sharing Act 2024 by notifying an individual that an automated decision-making process was employed, giving on request information about how it is employed, and providing a process for the individual to request human intervention in the decision.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.