{
 "public_id": "AIR-2025-0142",
 "incident_id": "2025-10-deloitte-australia-ai-report-refund",
 "date_disclosed": "2025-10-06",
 "date_occurred": "2025-07",
 "title": "Deloitte Australia refunded the federal government for a A$440k assurance report containing AI-fabricated citations and a fabricated court quote",
 "class": "safety_failure",
 "attack_surface": "other",
 "initial_vector": "model_hallucination",
 "agent_capability_abused": [
  "document_generation",
  "citation_synthesis"
 ],
 "target_product": "Azure OpenAI (GPT-4o) used to produce the Targeted Compliance Framework Assurance Review",
 "target_vendor": "Deloitte Australia (client: Australian Department of Employment and Workplace Relations)",
 "autonomy_level": "assistant",
 "impact_realized": true,
 "impact_type": [
  "financial_loss",
  "reputational_damage",
  "corrupted_deliverable"
 ],
 "blast_radius": "one government assurance review of the welfare compliance system; A$97,000+ refunded of a ~A$440,000 contract",
 "cve": [],
 "taxonomy": {
  "owasp_asi": [
   "ASI09"
  ],
  "mitre_atlas": [],
  "labels": [
   "fabricated_citations",
   "unverified_ai_output",
   "professional_deliverable"
  ]
 },
 "defense_present": "professional review process (failed to catch fabrications before publication)",
 "defense_bypassed": true,
 "fix_shipped": "revised report published with fabricated references and the judicial quote removed and an AI-use disclosure added; partial refund paid",
 "primary_sources": [
  "https://www.cfodive.com/news/deloitte-refunds-60k-report-ai-errors-australian-government-accounting/803321/",
  "https://oecd.ai/en/incidents/2025-10-05-be45"
 ],
 "verified_quote": "a fabricated quote from a federal court judgment and references to nonexistent academic research papers",
 "confidence": "high",
 "notes": "Boundary call. The model emitted text and called no tools, but the output was a delegated professional deliverable to a government client with a contractual and financial consequence, so it is not the 'text with no downstream effect' the scope note excludes. Errors were found by Chris Rudge (University of Sydney), including invented publications attributed to a real named academic. Contract executed December 2024, report finalised July 2025, errors reported early October 2025.",
 "risk_category": "brand_trust",
 "risk_secondary": [
  "security_fraud",
  "legal_ip"
 ],
 "risk_subcategory": "Reputation Harm via Misinformation",
 "risk_sub_secondary": [],
 "risk_override_reason": "fabricated citations in a paid government deliverable led to a refund and public reputational damage",
 "ai_entity": "generative",
 "agency": "human_acted_on_output",
 "agency_debatable": false,
 "agency_reason": "the model fabricated citations; Deloitte filed the report with government",
 "vector_family": "none_autonomous",
 "vector_detail": "model_hallucination",
 "guardrail_kind": [
  "abuse_detection"
 ],
 "guardrail_outcome": "silent",
 "xrefs": {
  "aiid": [],
  "atlas": [],
  "cve": [],
  "ghsa": []
 },
 "primary_asi": "ASI09",
 "agent_framework": "unknown",
 "model_provider": "multiple",
 "model_name": "",
 "detection_mechanism": "unknown",
 "tool_access": "unknown",
 "financial_loss_band": "unknown",
 "eai_category": "human_manipulation",
 "eai_domain": "agent_behaviour",
 "eai_secondary": [
  "unsafe_actuation"
 ],
 "eai_reason": "subcategory: Reputation Harm via Misinformation",
 "asi_curated": [],
 "asi_scope": "in_scope",
 "asi_provenance": "rule",
 "asi_reason": "ASI09 (eai_category=human_manipulation)",
 "submitted_by": "EAI Internal"
}