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 208

.au Domain Administration Ltd (auDA)

Submitter group: Other.

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.

auDA, the .au domain administrator, proposes that Australia recognise verifiable identity and accountability for autonomous AI agents as national digital infrastructure, and examine a sovereign, platform-neutral trust anchor built on the national DNS. Its white paper names auDA as a strong candidate operator, subject to an explicit mandate and oversight, and says it does not prescribe a technical implementation.

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 a national trust layer's purpose is to provide an authoritative, independently resolvable binding between an AI agent and the accountable entity that stands behind it.

    A national trust anchor should not attempt to determine whether an AI agent is safe, truthful or entitled to perform every action it requests. Its purpose is narrower and more fundamental: to provide an authoritative, independently resolvable binding between an agent and the accountable entity that stands behind it.

  2. Says systems accepting consequential AI-agent actions may need independent verification of the agent’s identity, accountable sponsor, current relationship and presented authority.

    Before accepting a consequential action from an AI agent, another system may need to establish independently what agent is acting, which person or organisation stands behind it, whether that relationship remains current, and where the authority claimed by the agent can be verified.

  3. Argues that if agent identity depends on proprietary platforms alone, the future internet risks fragmented namespaces, opaque accountability and higher switching costs.

    If durable agent identity becomes dependent on proprietary platforms alone, the future internet risks fragmented namespaces, opaque accountability and higher switching costs precisely when machines – not people – are making more of its decisions.

  4. Argues that Australia could retain sovereign control over agent trust infrastructure without building every foundation model or AI platform domestically.

    Australia need not develop every foundation model or AI platform domestically to retain sovereign control over important elements of the trust infrastructure through which autonomous agents interact with Australian organisations, citizens, government and critical systems.

  5. States that the proposal is not for DNS to decide whether an agent is safe or authorised, but to provide the public starting point for establishing accountability.

    The proposal is not that DNS should determine whether an agent is safe or authorised to undertake a particular transaction, but that it could provide the authoritative public starting point from which an agent's accountability and associated verification information can be established.

  6. Says auDA is a strong candidate operator because of its capability and public-interest role, subject to explicit conferral and transparent governance.

    In Australia, auDA is a strong candidate to operate that function because of its existing capability and public-interest role, but the function should be conferred explicitly and governed transparently.

Recommendations

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

  1. Recognise verifiable AI agent identity and accountability as an emerging national cyber security and digital infrastructure requirement.

    Recognise trusted, independently verifiable identity and accountability for AI agents as an emerging national cyber security and digital infrastructure requirement;

  2. Convene government, cyber security, digital identity, AI safety, critical infrastructure, privacy, competition and Internet-governance stakeholders to examine requirements for a sovereign, platform-neutral AI-agent trust anchor.

    Convene government, cyber security, digital identity, AI safety, critical infrastructure, privacy, competition and Internet-governance stakeholders to examine requirements for a sovereign, platform-neutral trust anchor for AI agents

  3. Support controlled technical trials and international standards engagement to test interoperable approaches before proprietary agent identity systems become entrenched.

    Support controlled technical trials and international standards engagement to test interoperable approaches before proprietary agent identity systems become entrenched.

  4. Retain government responsibility for policy, assurance requirements, sector obligations and oversight, with auDA potentially operating the narrower namespace and lifecycle service under an explicit mandate.

    Government should retain responsibility for policy, assurance requirements, sector obligations and oversight, while auDA could perform the narrower infrastructure role of operating the authoritative namespace and lifecycle service under an explicit mandate.

  5. Require national schemes in the proposed federation to publish machine-readable assurance semantics and recognition rules, since federation should not imply equivalent assurance.

    Because ccTLD governance and eligibility regimes differ, federation should not imply equivalent assurance: each national scheme should publish machine-readable assurance semantics and recognition rules.[15]

  6. Government, industry peak bodies and critical-infrastructure leaders should endorse a ccTLD-led registry, with its operator as lead convenor/intended operator, subject to transparent multi-stakeholder design, technical validation, competition safeguards and statutory/policy oversight.

    Government and industry should therefore recognise the ccTLD operator as the lead convenor and intended operator of the national AI agent trust namespace, subject to transparent multi-stakeholder design, technical validation and appropriate statutory or policy oversight. Recommended national action 1. Confer an explicit mandate. Government, industry peak bodies and critical-infrastructure leaders endorse a ccTLD-led national trust registry as shared digital infrastructure, subject to transparent multi-stakeholder design, competition safeguards and appropriate statutory or policy oversight.

  7. Establish a multi-stakeholder design group spanning cyber security, digital identity, AI safety, privacy, competition, law enforcement, procurement and standards.

    Establish a multi-stakeholder design group spanning cyber security, digital identity, AI safety, privacy, competition, law enforcement, procurement and standards.

  8. Pilot a 6–12-month sandbox with government services, banks, telecommunications, health, energy and major AI platforms; test DNSSEC resolution, organisational proofing, key rotation, bounded revocation latency, privacy controls, suspension and cross-platform verification.

    Pilot the architecture. Run a 6-12 month sandbox with government services, banks, telecommunications, health, energy and major AI platforms; test DNSSEC resolution, organisational proofing, key rotation, bounded revocation latency, privacy controls, suspension and cross-platform verification.

  9. Through procurement, require high-impact agents used by government or critical infrastructure to present a resolvable national identity and verifiable accountability chain, while accepting interoperable runtime credential schemes.

    Create demand through procurement. Require high-impact agents used by government or critical infrastructure to present a resolvable national identity and verifiable accountability chain, while accepting interoperable runtime credential schemes.

  10. Federate internationally through IETF processes, ICANN/IANA channels, regional ccTLD organisations and peer registries to explore common discovery, status and assurance semantics, treating current agent-identity drafts as work in progress.

    Federate internationally. Work through relevant IETF standards processes, Internet Corporation for Assigned Names and Numbers (ICANN)/ Internet Assigned Numbers Authority (IANA) coordination channels, regional ccTLD organisations and peer registries to explore common discovery, status and assurance semantics. Treat current agent-identity Internet-Drafts as work in progress rather than standards or institutional endorsement.

Read the original

Open in the interactive guide