Meta Mandates Weekly MetaCode Usage to Catch OpenAI Anthropic

Meta engineers ordered to submit weekly code diffs via MetaCode, aiming to accumulate training data and close the gap with OpenAI and Anthropic in AI coding.

Woofun AI reports that Meta has directed thousands of internal engineers to utilize its proprietary AI coding assistant, MetaCode, in an effort to enhance model capabilities and compete with OpenAI and Anthropic. Maher Saba, Meta's Vice President of Applied AI Engineering, issued an internal memo last month requiring engineers to complete at least one code "diff" per week using the tool. These reviewed modifications will be integrated to refine the AI models. Meta aims to gather extensive engineering practice data through real-world deployment, thereby improving performance in code generation, debugging, and software development. This initiative addresses the competitive pressure from industry leaders like Anthropic's Claude Code and OpenAI's Codex, as Meta seeks to narrow the capability gap through large-scale internal adoption.