The Fin 'Don’t do a Deloitte': the AI stuff-up that ricocheted around the world
25 snips
Oct 22, 2025 Edmund Tadros, a professional services editor, and Paul Karp, a political correspondent, dive into the global fallout from Deloitte's AI fiasco. They discuss how errors in a government report were uncovered, revealing the risks of relying on AI without proper oversight. The conversation highlights Deloitte's delayed admissions, the implications for consulting expertise, and the public's reaction to AI's potential threats. They also debate whether the incident is a cautionary tale for the industry or just the tip of the iceberg for future AI blunders.
AI Snips
Chapters
Transcript
Episode notes
How The Errors Were First Found
- Paul Karp discovered the Deloitte report's errors after a University of Sydney academic alerted him on August 22.
- He found more than 20 mistakes including fabricated quotes and nonexistent academic sources.
Big Fee For A Critical Code Review
- Deloitte was paid about $440,000 to review an automated welfare system and delivered the report in July.
- The report concluded the system might not work as intended and spurred departmental action before errors were flagged.
How Deloitte Responded And Fixed It
- Deloitte initially supplied an off‑record list of references and delayed taking the report down.
- The corrected version removed fake articles and rewrote the fabricated judge's quote.