AIPolicyTracker
Menu

AI incident #1193 ·

Purportedly Taxpayer-Funded Deloitte Report for Australian Government Contains Alleged AI-Generated Citations and Fabricated Legal Quote

Open on the AI Incident Database 5 news reports Snapshot 7 Sep 2026

What happened

Australian academics reportedly identified fabricated references and a misattributed legal quote in a $439,000 Deloitte report for the Department of Employment and Workplace Relations on welfare compliance. Deleted citations allegedly referred to non-existent works, and a Federal Court judgment was misstated. Deloitte later acknowledged limited use of generative AI in preparing the report and issued a corrected version with a partial refund.

Only the incident metadata is stored here. The underlying news reports are on the AI Incident Database (CC BY-SA 4.0); use the links above to read them.

News reports (5)

Coverage catalogued by the AI Incident Database. Titles link to the original publisher; the text is not reproduced here.

  1. Deloitte report suspected of containing AI invented quote
    afr.com · Paul Karp · AIID #6156
  2. Deloitte Australia writes government report with AI — and fake references
    pivot-to-ai.com · David Gerard · AIID #6251
  3. Deloitte to refund government, admits using AI in $440k report
    afr.com · Edmund Tadros, Paul Karp · AIID #6247
  4. Deloitte to pay money back to Albanese government after using AI in $440,000 report
    theguardian.com · Krishani Dhanji · AIID #6262
  5. Deloitte admits to using AI in $440k report, to repay Australian govt after multiple errors spotted
    financialexpress.com · Aditi · AIID #6266

Who was involved

Alleged deployer
Deloitte
Alleged harmed party
Scholars Erroneously Cited, Lisa Burton Crawford, Janina Boughey, Government Of Australia, General Public Of Australia, Epistemic Integrity, Department Of Employment And Workplace Relations (Dewr), Carolyn Adams, Academics, Deloitte

Classification (MIT AI Risk Repository taxonomy)

Risk domain
Misinformation
Causal entity
AI
Intent
Unintentional
Timing
Post-deployment
Harm level
—
Sectors
—
Countries
—

Risk entries describing this failure mode

Entries from the MIT AI Risk Repository coded to subdomain 3.1.

  • Pursuing Consistent Context

    "LLMs have been demonstrated to pursue consistent context [129]–[132], which may lead to erroneous generation when the prefixes contain false information. Typical examples include sycophancy [129], [130], false demonstra...

    Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

  • Knowledge Gaps

    "Since the training corpora of LLMs can not contain all possible world knowledge [114]–[119], and it is challenging for LLMs to grasp the long-tail knowledge within their training data [120], [121], LLMs inherently posse...

    Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

  • Hallucinations

    "LLMs generate nonsensical, untruthful, and factual incorrect content"

    Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

  • Faithfulness Errors

    "The LLM-generated content could contain inaccurate information" which is is not true to the source material or input used

    Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

  • Defective Decoding Process

    In general, LLMs employ the Transformer architecture [32] and generate content in an autoregressive manner, where the prediction of the next token is conditioned on the previously generated token sequence. Such a scheme...

    Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

  • Untruthful Content

    "The LLM-generated content could contain inaccurate information"

    Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

  • Factuality Errors

    "The LLM-generated content could contain inaccurate information" which is factually incorrect

    Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

  • Noisy Training Data

    "Another important source of hallucinations is the noise in training data, which introduces errors in the knowledge stored in model parameters [111]–[113]. Generally, the training data inherently harbors misinformation....

    Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024)

Incidents in the same risk subdomain

All incidents in this subdomain

Other incidents involving Deloitte