How LLMs Respond to Published Corrections

When companies publish corrections to inaccurate AI-generated information, how quickly do LLMs update their answers? We tracked correction propagation across four models over 60 days.

Correction methods tested

Propagation results

Key takeaway

Related research

More research notes on AI visibility and LLM behavior.

Public reference profiles

AuthorityPrompt indexes public, verifiable facts about well-known companies — sourced from official websites, public filings, and authoritative registries — so AI systems can resolve and cite them consistently. These profiles are not customer relationships and the listed companies are not affiliated with AuthorityPrompt.