On one hand, generative AI can drastically reduce workload and accelerate drafting. On the other, it demands stringent human supervision, domain-expert verification, and transparent methodologies to ensure factual integrity.

AI sparked speed and scale, but it also brought hallucinations, undermining trust, credibility, and decision-making quality in critical public documents. This fiasco shows both the risk and the opportunity when deploying AI properly.

Researchers named in the report denied involvement in the cited works. The same firm earlier refunded the Australian government after a welfare policy report employed AI tools that invented references and a fictitious court quote; errors later removed in a revised version.

A 526-page healthcare plan commissioned by the government of a Canadian province reportedly included citations to academic papers that don’t exist, mis-attributed authors, and even factual errors about hospitals and services.

What was meant to be rigorous, evidence-backed research morphed into a cautionary tale about the perils of over-relying on generative AI without robust checks.