“factual accuracy in large language models remains an area of active research”

This really says a lot.

Are they gonna file a complaint against Meta next because the Llama weights encode inaccurate facts on a person? File a complaint against the inference libraries too?