AI Inquiry Atlas

An unofficial, independent guide to the public records of the Joint Select Committee on Artificial Intelligence. Not connected with the Parliament of Australia or the committee. Its makers also made a submission (no. 481; see About). Summaries and scores are written by AI and can be wrong: check the original before relying on them.

Submission 412

Mr Andrew Ford

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.

Open in the interactive guide

Summary

Read the original submission for the full text.

Argues, in a personal capacity (the author is also a partner at a technology services firm advising government), that Australian laws are broadly adequate at the AI application layer but inadequate at the cloud and foundation-model layer, where public data and government dependency generate economic rent. Makes five recommendations; says its evidence is preliminary and the rent not yet quantified.

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.

  1. Argues that Australian law is broadly adequate for application-layer AI but structurally inadequate at the infrastructure layer.

    Australia’s laws are broadly adequate for AI at the application layer and are being incrementally adapted there. They are structurally inadequate at the infrastructure layer.

  2. Argues that no Australian law, regulator or policy instrument has the mandate, visibility or framework to identify cloud/AI providers’ economic rent from public data and public-sector dependency.

    No Australian law, regulator or policy instrument currently has the mandate, the visibility or the analytical framework to identify — let alone address — the economic rent that cloud and AI providers capture from public data and from public-sector dependency.

  3. Argues that no Australian law treats public data as a national asset whose commercial AI use should generate a return to the public.

    There is, in short, no Australian law that treats public data as a national asset whose use in commercial AI should generate a return to the public, in the way that mineral resources attract royalties

  4. Says government cloud dependency is treated as procurement rather than regulation, with whole-of-government arrangements, single-seller arrangements and cloud-first policy designed to lower price and friction.

    Government cloud dependency is treated as a procurement matter, not a regulatory one. Whole- of-government arrangements, single seller arrangements and the 2026 cloud-first policy have been designed to lower price and friction.

  5. Argues that diffuse harm to public value escapes harms registers, requiring broader gap-analysis terms addressing market structure, dependency and public value return.

    The harm is to public value and to national capability, and it is diffuse, cumulative and invisible to a harms register. Unless the terms of reference for the gap analysis are broadened to include market structure, dependency and public value return, the analysis will confirm the adequacy of frameworks that were never asked the right question.

  6. Says Australia lacks equivalents to EU cloud-switching-charge abolition and UK voluntary egress/interoperability commitments for business or government customers.

    The EU Data Act abolishes cloud switching charges from 12 January 2027 and the UK has extracted (voluntary) egress and interoperability commitments from AWS and Microsoft. Australia has no equivalent for business or government customers.

Recommendations

Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.

  1. Prioritise cloud and foundation-model designation alongside app marketplaces and ad tech when legislating the competition regime; include Commonwealth workload/spend dependency qualitatively and compulsory information powers over training-data provenance and tenant workloads.

    Recommend that cloud infrastructure and foundation-model services be a priority designation stream alongside app marketplaces and ad tech when the regime is legislated, with designation criteria that include public-sector dependency (share of Commonwealth workloads and spend) as a qualitative factor, and with compulsory information-gathering powers extending to model-training data provenance and the use of tenant workloads.

  2. Legislate a statutory scheme modelled on Chapter VI of the EU Data Act (abolishing switching charges, mandating exit assistance and open interfaces) for all cloud customers, with rights non-waivable in government contracts.

    Recommend a statutory scheme modelled on Chapter VI of the EU Data Act — abolition of switching charges, mandated exit assistance and open interfaces — applying to all cloud customers, with a Commonwealth-specific schedule that makes these rights non-waivable in government contracts.

  3. Include public returns from commercial Commonwealth-data AI use in the 2027 DAT Act and Attorney-General’s training-data framework: licence conditions, revenue-linked Australian AI capability fund contributions, or model access for Australian public institutions.

    Recommend that the re-enacted DAT Act (2027) and the AI training-data licensing framework under development by the Attorney-General’s Department include an express principle that commercial AI use of Commonwealth data attracts a return to the public — through licence conditions, revenue-linked contributions to an Australian AI capability fund, or mandated model access for Australian public institutions — and that open-data licensing policy be reviewed against the same principle.

  4. Review open-data licensing policy against the public-value-return principle.

    Recommend that the re-enacted DAT Act (2027) and the AI training-data licensing framework under development by the Attorney-General’s Department include an express principle that commercial AI use of Commonwealth data attracts a return to the public — through licence conditions, revenue-linked contributions to an Australian AI capability fund, or mandated model access for Australian public institutions — and that open-data licensing policy be reviewed against the same principle.

  5. Require annual Office of AI, ACCC and DTA reports to Parliament on Commonwealth cloud/model dependency: provider spend overriding aggregate-disclosure confidentiality restrictions, workload concentration, estimated switching costs, and government-data/telemetry-use contractual terms.

    Recommend that the Office of AI, the ACCC and the DTA report annually to Parliament on Commonwealth cloud and foundation-model dependency: spend by provider (overriding contractual confidentiality restrictions on aggregate disclosure), workload concentration, estimated switching costs, and the contractual terms governing provider use of government data and telemetry.

  6. Extend National AI Plan gap analysis to infrastructure market structure, sovereign dependency and public value capture, drawing on ACCC, Productivity Commission and independent academic expertise; publish findings.

    Recommend that the gap analysis committed to in the Plan be expressly directed to infrastructure-layer market structure, sovereign dependency and public value capture, drawing on the ACCC, the Productivity Commission and independent academic expertise, and that its findings be published.

Read the original

Open in the interactive guide