Meta Platforms Inc Launches New Major Language Model And Remains Committed To Open AI Research

JAKARTA - Meta Platforms Inc announced on Monday, March 6 that they will continue to release their artificial intelligence (AI) tools to approved researchers despite claims in online forums that the latest major language model has been leaked to unauthorized users.

"While this model is not accessible to everyone, and some have tried to circumvent the approval process, we believe that the current release strategy allows us to maintain a balance of responsibility and openness," Meta said in a statement, quoted by Reuters.

Facebook owns Meta, maintains a large AI research division and last month released LLaMA, short for Meta AI Big Language Model. Meta claims that the model can achieve human-like conversational capabilities from AI systems created by the creators of ChatGPT, OpenAI, and Alphabet Inc, but only uses far less computing power.

Unlike some of its competitors such as OpenAI, which severely restricts its technology and charges software developers access to it, AI Meta's research division shares much of its work publicly. However, these AI tools also have the potential for abuse, such as creating and spreading false information.

To avoid this type of abuse, Meta makes its tools available to researchers and other entities affiliated with government, civil society, and universities under a non-commercial license after going through a selection process.

Last week, users on the online forum 4Chan claimed to have made this model available for download. Reuters was unable to independently verify the claim.

In its statement, Meta said that their LLaMA release is being handled in the same way as previous models and that they have no plans to change their strategy.

"Meta's goal is to share the latest AI models with members of the research community to help us evaluate and improve those models," said Meta.