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 267

National Employment Services Association (NESA)

Submitter group: Industry bodies.

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.

Peak body for employment services providers argues that case-by-case approval of AI use by providers is slow and opaque while administrative burden is high. Recommends an AI enablement framework with risk tiers, a fast pathway for low-risk uses, co-designed pilots before November 2028, human oversight of consequential decisions, funding for adoption, and shared governance.

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. States that initial workforce-survey findings show a median 50% of frontline and manager time goes to administration and compliance.

    Skilled practitioners are spending half their time on administration. Initial findings from NESA's 2025–26 Workforce Survey2 conducted by the Centre for International Economics across 142 employment services providers Australia wide, show a median 50% of frontline and manager time goes to administration and compliance.

  2. States that, as at 17 September, there are no published criteria, decision timeframes or pre-approved low-risk uses.

    As at 17 September, there are no published criteria, decision timeframes or pre-approved low-risk uses.

  3. Argues that the Commonwealth is enabling its own workforce with AI while holding back its delivery partners.

    The Commonwealth is enabling its own workforce while holding back its delivery partners.

  4. Argues that delay shifts risk, because unmanaged use of public AI tools is the likely alternative to approved tools.

    Where approved, governed tools are unavailable, the likely alternative is unmanaged individual use of public tools.

  5. States that AI tools must complement, not replace, the human relationship between practitioners and job seekers.

    These tools must complement, not replace, the trusted human relationship that evidence shows drives outcomes for people with complex barriers.

Recommendations

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

  1. DEWR, DSS, NIAA and the DTA should publish within six months a sector-specific, risk-based framework setting out permitted, conditional and prohibited uses across employment services programs.

    The Department of Employment and Workplace Relations (DEWR), the Department of Social Services (DSS), National Indigenous Australians Agency (NIAA) and the Digital Transformation Agency, should publish within six months a sector-specific, risk-based framework. It should set out permitted, conditional and prohibited uses across Workforce Australia, Inclusive Employment Australia, the Remote Australia Employment Service and related programs.

  2. Recognise one approval across all three agencies named in the framework recommendation.

    One approval should be recognised by all three agencies.

  3. DEWR should publish assessment criteria and decision timeframes and pre-approve low-risk productivity tools, as well as consent-based transcription, summarisation and draft case-note preparation.

    DEWR should publish assessment criteria and decision timeframes, and pre-approve low-risk productivity improvement tools, as well as consent-based transcription, summarisation and draft case-note preparation.

  4. Benchmark decisions at within 30 business days for standard applications, with all outstanding applications resolved by the end of 2026.

    NESA recommends benchmarking for decisions to be made within 30 business days for standard applications, with all outstanding applications resolved by the end of 2026.

  5. Fund co-designed pilots before November 2028 across provider sizes and metropolitan, regional, remote, disability and First Nations settings, independently evaluating outcomes, productivity, accessibility, safety and workforce effects and publishing results.

    Fund co-designed pilots before the new system commences in November 2028. The Government should fund pilots designed with job seekers, frontline practitioners, providers and technology experts. Pilots should span provider sizes and metropolitan, regional, remote, disability and First Nations settings. They should be independently evaluated for participant outcomes, productivity, accessibility, safety and workforce effects, with results published.

  6. Require program Deeds and guidelines to ensure meaningful human oversight and named accountability for decisions affecting eligibility, payments, compliance, service allocation or other rights and interests, equally for departmental and provider systems.

    The Government should require, in program Deeds and guidelines, meaningful human oversight and named accountability for any decision affecting eligibility, payments, compliance, service allocation or other rights and interests. This should apply equally to departmental and provider systems and responsibilities.

  7. Recognise AI implementation, training and secure data costs in the new employment services funding model and across programs, and make departmental systems interoperable to reduce duplicate entry.

    The Government should recognise AI implementation, training and secure data costs in the new employment services funding model and across all employment services’ programs and make departmental systems interoperable to reduce duplicate entry.

    A program corrected obvious text-recognition errors in this quote for display.

  8. Give smaller, regional and First Nations providers access to shared, pre-assured tools, with capability uplift within the national employment services Workforce Plan NESA has proposed.

    Smaller, regional and First Nations providers should have access to shared, pre-assured tools. Capability uplift should sit within the national employment services Workforce Plan NESA has proposed under the employment services reforms.

  9. DEWR should convene a standing AI working group with DSS, NIAA, providers, participant representatives and NESA, to update guidance, share use cases, identify risks and report annually on effects.

    DEWR should convene a standing government-employment services sector AI working group with DSS, NIAA, providers, participant representatives and NESA. It should update guidance, share approved use cases, identify emerging risks, and report annually on effects for job seekers, employers and frontline work.

    A program corrected obvious text-recognition errors in this quote for display.

  10. Apply safeguards proportionately to risk, distinguishing practitioner-checked case-note drafting from automated decisions affecting income support.

    Safeguards must also be proportionate to risk. Drafting a case-note summary for a practitioner to check should not be regulated like an automated decision capable of affecting a person's income support. NESA proposes three tiers.

  11. Pre-approve low-risk drafting, summarising and note-taking in enterprise tools within providers’ secure environments against published safeguards, with provider self-attestation and notification.

    Low risk Drafting, summarising and Pre-approved against note-taking in enterprise tools | published safeguards; held within the provider's provider self-attests and secure environment notifies

  12. For conditional uses, require standard applications, privacy impact assessments, defined human review, audit logging and published decision timeframes for recording/case-note generation, matching, participant coaching and departmental-system integration.

    Conditional Consent-based recording and | Standard application, privacy case-note generation, job impact assessment, defined matching, participant-facing human review, audit logging, coaching tools, integration published decision with departmental systems timeframes

  13. Prohibit fully automated decisions on eligibility, streaming, payments, compliance, mutual obligations or sanctions; offshore participant-data export; and training third-party models on participant data.

    Prohibited Fully automated decisions on | Not permitted, and clearly eligibility, streaming, listed payments, compliance, mutual obligations or sanctions; offshore export of 12 Royal Commission into the Robodebt Scheme, Report, July 2023, recommendations 17.1 and 17.2 on automated decision-making. 13 NESA, Submission on Shaping the future of Employment Services, July 2026 Tier Examples Treatment participant data; training third-party models on participant data

  14. Across tiers, require named accountability; informed plain-language recording consent and a non-AI alternative; AI-use transparency; data minimisation, purpose limits, Australian storage, access controls and audit trails; fairness, accessibility and cultural safeguards; and independent evaluation.

    human accountability: a named person responsible for every consequential decision consent and choice: informed, plain-language consent before conversations are recorded, and a genuine non-AI alternative transparency: telling people when and how AI is used, consistent with the Privacy Act 1988 transparency obligations for automated decisions that commence on 10 December 2026 data protection: data minimisation, purpose limitation, Australian data storage, access controls and audit trails fairness and safety: testing for bias and accessibility, culturally safe design with First Nations communities, and added protections for young people and vulnerable participants, and monitoring: regular review and independent evaluation.

    A program corrected obvious text-recognition errors in this quote for display.

  15. Preserve face-to-face, practitioner-led assistance so AI widens rather than restricts access to support.

    Face-to-face, practitioner-led assistance must be preserved. With one in five Australians being digitally excluded, rising to 41% of First Nations people, with the gap widest in remote areas, AI must widen access to support, not become a new barrier to it.

    A program corrected obvious text-recognition errors in this quote for display.

Read the original

In some quotes a program corrected obvious text-recognition errors for display.

Open in the interactive guide