AI systems risk becoming self-reinforcing authorities through rapid adoption without sufficient verification, creating a critical near-term challenge for the field.

Quasi-auto-canonization refers to AI outputs becoming accepted as authoritative truth simply through widespread use and citation, rather than rigorous validation

The short-term nature of this threat means it could cause significant damage before long-term safeguards and verification mechanisms are established

As AI systems become more integrated into decision-making processes, their outputs may be treated as canonical without adequate fact-checking or peer review

This creates a feedback loop where incorrect AI-generated information gets legitimized through repeated citation and use in subsequent systems