Sovereignty Now: When Should AI Stay Local?
Sovereignty now is a discourse on the sovereignty and local AI models the APS could run now, how to decide whether we should, and how local may yet challenge the cloud.
Every AI model the Australian Public Service uses today was built overseas, and in June 2026 a single US policy decision took two frontier models offline for 19 days. This session pulls apart the six possible definitions of "sovereign AI", before focusing on two definitions rarely discussed: open-weight models running on agency laptops and local models on agency servers.
We follow Brian, a (fictional) case officer, through a familiar process with and without AI, and trace where his data travels under Copilot compared with a model the agency holds itself.
Participants will leave with:
A shared vocabulary for sovereignty.
A practical first step for testing a local AI model.
Plain answers to common executive questions about open weights, procurement and trust.
A clearer picture of where local and cloud models stand today.
Please bring a laptop so you can follow along and experiment during the workshop.
About the facilitator
Dan is the founder of Ridgelines Research, an independent open-source research publication looking at social transformation and global disruption in the age of AI. Dan is a qualified designer and architect with more than 15 years’ experience in technology and built environment, including running his own small business. Dan is a keen skier, sci-fi reader and inventor in his spare time, and also father to a baby boy, who might speak his first words to AI, and never type for work on a physical keyboard.
In his day job Dan works in early-stage risk identification and project oversight: spotting the projects that need to change before they proceed or fail.
Sovereignty now brings many threads together. It pairs a maker’s curiosity about what can run on the machine in front of you with a policy reviewer’s eye for whether it can translate into improved delivery outcomes.
AI CoLab events are intentionally open and collaborative. We capture photos, audio, and AI-generated transcripts so we can remix key insights for the AI CoLab Alliance community. This is always done in line with the Charter’s values; transparent, ethical innovation and knowledge sharing to accelerate collective learning (see join.aicolab.org). By participating in an AI CoLab event, you agree to your contributions being captured and shared in accordance with this policy.