Like many other not-for-profit organisations, First Step has just completed its first policy on the use of artificial intelligence.
It was a significant piece of work, involving contributions from across the organisation , and guided by external references such as the Australian Health Practitioner Regulation Agency and the Royal Australian College of General Practitioners.
One of the first insights was that staff needed a baseline of AI education before any policy could be embedded. We needed clarity on the differences between descriptive tools, generative tools, licensed platforms, and unlicensed platforms. Just as important, we recognised that the implications of AI vary greatly across our work. For that reason, the policy has been designed around three distinct areas: clinical, legal, and administrative.
At its core, the policy is anchored in principles that apply universally, regardless of technology.
These are: safe (prioritising client well-being), transparent (AI use is clearly identified), confidential (data security and privacy are protected), human-led (AI supplements but never replaces human judgment), and equitable (avoiding bias and discrimination).
The policy also sets out permitted and prohibited uses, with clear processes for each area.
Permitted use includes drafting or summarising clinical notes, but only WITH human review and only WITHIN our licensed client data management system. That clarity has helped staff understand where AI adds value and where it must not intrude.
A prohibited use is uploading any personal information about a client on to a public platform, even if the information is completely deidentified.
A procedural example is always asking permission from a medical patient before using an A.I. Scribe. It might be expected and standard one day, but not yet.
Finally, unlike most policies, this one is deliberately iterative. We expect to review and update it every few months as both the technology and our understanding evolve.
It’s a challenging but exciting future, and we are ready to face it with clear values and practical safeguards.
And as for the last principle, equitable (avoiding bias and discrimination), we expect this will be the most important over time.
