Updated
Updated · The Guardian · Sep 2
Australian Committees Warn AI Hallucinations Are Flooding Inquiries, Citing a $440,000 Fake-Reference Report
Updated
Updated · The Guardian · Sep 2

Australian Committees Warn AI Hallucinations Are Flooding Inquiries, Citing a $440,000 Fake-Reference Report

3 articles · Updated · The Guardian · Sep 2

Summary

  • Australian parliamentary committee chairs said AI-generated submissions are feeding false and even hallucinated material into federal inquiries, raising the risk that MPs make policy decisions on incorrect information.
  • Guardian Australia found some inquiry submissions invented studies and misattributed research to real academics, and some committee reports cited submissions whose sources appeared largely AI-generated.
  • Paul Scarr said committees must consider all submissions but now need to check footnotes, sources and claims more closely; he sat on a 2023 inquiry that received a submission containing fake allegations against KPMG.
  • Louise Miller-Frost said committees and secretariats are more alert to the problem, though heavy submission volumes make every reference hard to verify and AI-assisted submissions should not be automatically discarded.
  • Sue Lines said submitters remain responsible for accuracy as committee practices evolve; the warning follows a 2025 Deloitte partial refund after AI inserted fake references, including a bogus court citation, into a $440,000 government report.

Insights

Will cracking down on AI submissions silence the marginalized voices who rely on them?
Could fake AI data already be secretly shaping national laws without anyone noticing?