Submission 246
Dr Jamie Freestone
Submitter group: Individuals.
Position labels are not shown for individual submitters.
An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.
Summary
Read the original submission for the full text.
An AI consultant and researcher argues that AI makes state data integration far cheaper, without an equivalent gain for citizens, while existing controls focus on collection. States that the submission makes no accusation of misconduct. Recommends oversight resourcing, capability reporting, reauthorisation and procurement exit tests, without restricting what agencies may acquire.
Key points
Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.
Argues that an agency’s analytic capability can increase by an order of magnitude without new legislation, authorisation or a recorded decision, because nothing new was collected.
Hence the central sentence of this submission: an agency’s analytic capability can increase by an order of magnitude with no new legislation, no authorisation, no threshold crossed and no decision appearing on any record — because nothing new was collected. Every control we operate watches the collection boundary.
Argues that the cost of joining datasets historically acted as a control on which accountability arrangements relied.
Historically that joining was rationed by cost, so it happened only where someone senior decided it was worth doing. Scarcity was functioning as a control, and our accountability architecture quietly relied on it.
States that the author could not find publicly available information on the Commonwealth’s purchased integrated analytic capability.
I cannot tell this Committee what integrated analytic capability the Commonwealth has purchased, or what those systems can do across combined holdings. I have looked; it is not publicly available, and so far as I can determine the Committee cannot readily find out either.
States that at 30 June 2025 IGIS was operating at an average staffing level of 43.48 against an approved 68.5, across six agencies.
At 30 June 2025 the Inspector-General of Intelligence and Security was operating at an average staffing level of 43.48 against an approved 68.5, across six agencies.
Argues that integration platforms absorb schemas, workflows and institutional memory, with rising exit costs narrowing the Commonwealth’s window of genuine choice.
Integration platforms absorb data schemas, workflows and institutional memory, and exit costs rise every month, so the window in which the Commonwealth retains a genuine choice is far shorter than the life of the relationship.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Find that AI's effect on the state-citizen relationship is asymmetric: materially increasing the state's capacity to know about citizens with no equivalent increase in citizens' capacity to inform and organise.
That the Committee find AI’s effect on the state–citizen relationship to be asymmetric: materially increasing the state’s capacity to know about citizens with no equivalent increase in citizens’ capacity to inform and organise themselves.
Resource IGIS and the INSLM to approved staffing levels and fund IGIS for a standing technical inspection capability with vendor-independent access to interrogate integrated analytic systems directly.
That IGIS and the INSLM be resourced to their approved staffing levels, and IGIS additionally funded for a standing technical inspection capability: personnel, tooling and vendor-independent access sufficient to interrogate integrated analytic systems directly rather than through agency briefings.
Require agencies to report acquired integrated analytic capability, and how it has changed, rather than enumerating tools or use cases.
That agencies report acquired integrated analytic capability: what their systems can now do across combined holdings and how that has changed since last reporting, rather than enumerating tools or use cases.
Apply automated and AI-assisted decision-making arrangements to intelligence and security agencies through a classified channel reporting to IGIS, the INSLM and the PJCIS, rather than by exemption.
That arrangements for automated and AI-assisted decision-making apply to intelligence and security agencies through a classified channel reporting to IGIS, the INSLM and the PJCIS, rather than by exemption.
Subject integrated data-analytic platforms operated by security and law enforcement agencies to periodic reauthorisation, so continuation requires a positive decision.
That integrated data-analytic platforms operated by security and law enforcement agencies be subject to periodic reauthorisation, so continuation requires a positive decision rather than administrative inertia.
Require security and law enforcement AI and data-integration procurement to pass a pre-contract exit test: escrow of models, schemas and audit logs, demonstrated reconstruction of outputs without vendor help, and a costed, time-boxed exit path.
That AI and data-integration procurement by security and law enforcement agencies face an explicit pre-contract exit test: escrow of models, schemas and audit logs; demonstrated Commonwealth ability to reconstruct any analytic output without vendor assistance; and a costed, time-boxed exit path.