Most AI governance in Australia is built on policy documents and good intentions. We build the evidence that holds when a regulator asks.
Independent AI safety research at UTS. We study what AI systems actually do — not what their developers claim they do. No commercial agenda. Peer-reviewed. The research foundation everything else is built on.
Sam works with four organisations at any given time. Not as a vendor — as the person accountable for your AI governance program. Your board is asking questions. This is where you get answers.
Built on the MIT AI Risk Repository — 1,612 risks, 65 taxonomies, mapped to your systems and your Australian legal obligations. Board-ready governance evidence in 48 hours. No consultant required.
AI Decoded was founded by Sam B and Rae D — two practitioners who got tired of the same answer. Sam brings PhD research at UTS, the CausalShield institute, and four companies built and exited. Rae brings 18 years of strategic market intelligence across fintech and regulated industries, and built the operational framework that turns the MIT AI Risk Repository into evidence a board can actually use. The practice exists because most AI governance in Australia is built on opinion, vendor marketing, and frameworks that nobody has operationalised.
Every claim we make is anchored in independent research. Every product we build is grounded in four independent evidence bases — the MIT AI Risk Repository, the Privacy Act 1988 APP obligations, the CSET AI Harm Taxonomy, and the AIID Incident Database. Every advisory engagement produces documentation defensible on the day a regulator asks for it.
"The 2026 regulatory theme is proof, not promises — regulators expect auditable evidence that compliance controls work in practice, not just policies on a shelf."
LexisNexis · Legal Year in Review 2026