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 51

Workday

Submitter group: Companies.

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.

An enterprise software vendor argues for open competition and technology choice in AI sovereignty, skills-based workforce development and governance by design in government AI. It opposes a standalone AI Act while supporting targeted mandatory high-risk safeguards alongside voluntary general standards. Its supplementary submission proposes detailed safeguards for sensitive enterprise agents and alignment with international frameworks.

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 sovereignty comes from open competition, technology choice and access to global AI tools.

    Rather, genuine sovereignty comes from open market competition, technological choice, and the ability of Australian businesses and government agencies to access the world’s most secure, performant, and value-driven AI tools.

  2. Describes a skills-based workforce approach focused on individual capabilities and organisational needs rather than credentials alone.

    A skills-based approach focuses on the specific capabilities an individual possesses and those an organisation needs, rather than relying solely on credentials, qualifications, or job titles.

  3. Argues that transparency and accountability must be built into AI deployment by design.

    Workday firmly believes that transparency and accountability must be built into AI deployment by design, not bolted on as an afterthought.

  4. Recommends maintaining voluntary, technology-neutral standards for general AI uses, combined with targeted mandatory safeguards for high-risk deployments.

    Workday supports maintaining Australia’s current technology-neutral, voluntary-standards approach for general AI uses, combined with targeted, mandatory safeguards for high-risk deployments.

  5. Supplementary submission recommends treating agentic AI governance as a distinct but complementary element alongside frontier model safety.

    Treat agentic AI governance as a distinct, but complementary, element of AI governance alongside frontier model safety — reflecting that the risks of ungoverned enterprise agents are different in kind from the risks raised by frontier model training and evaluation activity.

Recommendations

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

  1. Build national resilience through open competition, global innovation and technological choice rather than protectionist mandates or restrictive vendor pools.

    Build national resilience through open competition, global innovation, and technological choice rather than protectionist mandates or restrictive vendor pools.

  2. Keep Australian businesses and government agencies free to procure enterprise technologies globally.

    Ensuring Australian commercial enterprises and government agencies remain free to procure best-of-breed enterprise technologies globally, avoiding technological isolation and empowering the Australian workforce with peak operational capabilities.

  3. Foster vendor competition on security, performance and ethical AI standards.

    Fostering a dynamic vendor ecosystem where global enterprise software providers compete transparently on security benchmarks, system performance, and ethical AI standards.

  4. Enforce strict data governance, onshore data residency capabilities for highly sensitive data, and robust security frameworks safeguarding Australian data wherever software models are engineered.

    Protecting national interests by enforcing strict data governance, onshore data residency capabilities for highly sensitive data, and robust security frameworks that safeguard Australian data regardless of where software models are engineered.

  5. Shift towards a skills-based approach to workforce development.

    In an economy where digital literacy and an understanding of AI is becoming a requirement for most roles, a transition towards a skill-based approach is essential for promoting development and future-proofing the workforce.

  6. Develop and integrate the National Skills Taxonomy as the common language for understanding skills across the economy.

    The Government would gain significant benefits from the development and integration of the National Skills Taxonomy (being developed by Jobs and Skills Australia) as the common language for understanding skills across the entire economy.

  7. Improve access to accurate, updated labour-market data as a critical step toward skills-based workforce development.

    To achieve the benefits of a skills-based approach to workforce development, a critical step would require better access to labour market data that is accurate and updated.

  8. Improve access to labour market data and how it is collected and used, to build skills intelligence capabilities.

    building skills intelligence capabilities requires not just improved access to labour market data but also how such data is collected and used.

  9. Combine comprehensive labour-market data access with advanced skills intelligence, including predictive AI planning, to inform policy and workforce planning.

    Unlocking access to comprehensive labour market data, combined with advanced skills intelligence capabilities – such as moving from static reporting to predictive, AI-driven foresights - offers a powerful mechanism to address these issues, inform policy, and enhance planning for a skilled and adaptable workforce.

  10. Ensure workers understand when and how AI materially informs decisions affecting them, including hiring, performance management and role changes, consistently with human-oversight principles.

    we encourage a focus on transparency: workers should be able to understand when and how AI materially informs a decision that affects them (such as hiring, performance management, or role changes), consistent with the human-oversight principles Workday embeds in its own Responsible AI framework.

  11. Modernise legacy infrastructure and operationalise governance-by-design so agencies can scale trusted AI.

    Modernise legacy infrastructure and operationalise governance-by-design to give Commonwealth agencies the confidence to scale trusted AI solutions.

  12. Ensure the Government's AI governance frameworks are actively operationalised across all agencies, so AI can move from isolated pilots into core operational systems.

    We encourage the Government to build on this momentum by ensuring these frameworks are actively operationalised across all agencies, giving public sector leaders the confidence to transition AI from isolated pilots into core, trusted operational systems.

  13. Preserve the technology-neutral regulatory foundation and introduce proportionate, risk-based guardrails for high-risk deployments rather than enacting a standalone AI Act.

    Preserve Australia's technology-neutral regulatory foundation and introduce proportionate, risk-based guardrails for high-risk deployments rather than enacting a standalone AI Act.

  14. Explicitly distinguish the regulatory roles of AI developers and AI deployers.

    Regulatory responsibilities must explicitly distinguish between the roles of AI developers (who build models) and AI deployers (who operationalise them in specific context settings).

  15. Use context-aware impact assessments for risk management; defer mandatory third-party audits as premature, while allowing reconsideration as global assurance frameworks mature.

    Accountable governance tools like impact assessments provide a practical, context-aware mechanism for risk management. In contrast, mandatory third-party auditing requirements remain premature given the current maturity of technical standards and market oversight infrastructure, though this can be revisited as global assurance frameworks mature.

  16. Finalise a clear controller-processor distinction in the Privacy Act reforms, with due-diligence and binding compliance agreements.

    Modernising the Privacy Act remains a foundational pillar for governing AI data inputs and handling. In particular, we strongly encourage the finalisation of a clear controller–processor distinction alongside accompanying due-diligence and binding compliance agreements.

  17. Maintain structured, ongoing dialogue with industry on how existing technology-neutral frameworks, including privacy, WHS and anti-discrimination laws, can be calibrated alongside tailored safeguards.

    We encourage the Committee and Government to maintain structured, ongoing dialogue with industry to evaluate how existing technology-neutral frameworks—including privacy, work health and safety, and anti-discrimination laws—can be calibrated dynamically alongside tailored safeguards as high-risk use cases emerge.

  18. Treat agentic AI governance as a distinct but complementary element of AI governance alongside frontier model safety.

    Treat agentic AI governance as a distinct, but complementary, element of AI governance alongside frontier model safety — reflecting that the risks of ungoverned enterprise agents are different in kind from the risks raised by frontier model training and evaluation activity.

  19. Apply safeguards to sensitive enterprise agents capable of state-changing HR or finance actions that could significantly affect employment, personal information, or enterprise records, assets, obligations or controls.

    Workday recommends that safeguards apply to “sensitive enterprise agents”, i.e. agents that are capable of executing state-changing actions in HR or finance systems that could have significant effects on a person’s employment or personal information, or an enterprise’s records, assets, obligations, or controls.

  20. Require each sensitive enterprise agent to have persistent non-human identity, documented purpose and authority, and an auditable permission record in a central registry.

    Every sensitive enterprise agent should have a persistent, non- human identity, a documented purpose and scope of authority, and an auditable permission record, retained in a central registry.

  21. Require a designated accountable manager, defined decision rights and escalation paths, and human mechanisms to review, approve, override or halt each agent.

    Each agent should have a designated, accountable manager on record, defined decision rights and escalation paths, and mechanisms for humans to review, approve, override, or halt its actions.

  22. Require data minimisation and least-privilege access across internal permissions, external connections and tool calls, including through open standards such as Model Context Protocol and agent-to-agent communication.

    Agents should enforce strict data minimisation and least-privilege access across internal permissions, external system connections, and tool-calling capabilities, including through open standards such as Model Context Protocol and agent-to-agent communication.

  23. Require trace logs sufficient to reconstruct agent actions, the principal, authority, systems used and any human approval, override or escalation.

    Agents should maintain trace logging sufficient to reconstruct and audit what an agent did, on whose behalf, under what authority, using which systems, and whether human approval, override, or escalation occurred.

  24. Evaluate agents against recognised standards before deployment, document assurance, and require independent third-party evaluation for agents delegated significant discretion over high-stakes decisions.

    Agents should be evaluated against recognised standards and frameworks before deployment, documented in an assurance record — with independent third-party evaluation for agents delegated significant discretion over high-stakes decisions.

  25. Adopt policies advancing five enterprise-agent safeguards (persistent identity and authority, human oversight, data protection and security, auditable logs, assurance before deployment) as a checklist for AI-agent risk in Commonwealth adoption.

    Adopt policies that advance the five enterprise-agent safeguards — persistent identity and authority, rigorous human oversight, data protection and security, auditable logs, and assurance prior to deployment — as a practical checklist for evaluating AI-agent risk in Commonwealth department and agency adoption.

  26. Preserve clear delineation of developer and deployer roles across the agentic AI value chain.

    Preserve clear delineation of roles and responsibilities across the agentic AI value chain, tailoring obligations to developers and deployers according to their distinct visibility and control.

  27. Anchor any future Australian AI standards or legislation in NIST AI RMF and ISO/IEC 42001.

    Anchor any future Australian standards for AI, or legislative response, in established international frameworks, in particular the NIST AI Risk Management Framework and ISO/IEC 42001, to maintain consistency with Australia's trading partners and avoid conflicting cross-border obligations for industry.

Read the original

Open in the interactive guide