Meta has banned its Applied AI engineers from using Anthropic's Claude Code and OpenAI's Codex, citing concerns over model distillation, data contamination, and tech leakage. This reflects a growing industry battle where competitors' outputs are used to train rival models. Since 2026, accusations of distillation violations have escalated among major AI firms. Meta's move signals a shift from open-source advocacy to building a data moat, and may prompt other players to enforce similar restrictions.