In the field of large language models, Meta's Galactica model was taken offline just three days after its release due to issues with hallucinations. The model was considered an early attempt and had a positive impact on Meta's subsequent projects. At the time of its release, the foundational model was not thoroughly checked due to a shortage of personnel, but the model's first author, Ross Taylor, emphasized that the research still holds value. Joelle Pineau, Vice President of AI Research at Meta, stated that Galactica was essentially a research project. The model's shutdown drew attention from individuals like Yann LeCun, who warned that overemphasizing AI ethics could have counterproductive effects.