On February 14, 2019, OpenAI announced via a blog post that it would not release the full algorithm for its text-generation model GPT-2, citing "safety and security concerns."
Instead, OpenAI released only a "much smaller" version of the model, keeping the dataset and training code used for its development private.
GPT-2 was developed as a model capable of generating coherent and diverse text when given prompts on specific topics. OpenAI is a nonprofit research organization initially funded by Elon Musk, Peter Thiel, Reid Hoffman, and others, with a mission to guide AI development in a responsible direction.
Following this announcement, discussions arose among machine learning experts questioning whether OpenAI's claims were somewhat exaggerated. The announcement also sparked debate on how to handle the proliferation of AI algorithms with potentially dangerous capabilities.
Source: OpenAI says its new model GPT-2 is too dangerous to release (2019) (HN 395pt・120 comments) (HN Search (backfill))