$META is limiting how engineers use Anthropic’s Claude Code and OpenAI’s Codex for AI model-building work, per The Information.
Internal docs show Meta is concerned rival model outputs could leak into its own training data and trigger distillation issues.