Meta has quietly moved to restrict its engineers from using two of the most popular AI coding agents on the market: Anthropic’s Claude Code and OpenAI’s Codex. The policy, which appears to be in effect as of June 2026, is aimed at protecting the company’s proprietary AI training data and internal codebases from potential exposure.
What Meta is actually worried about
Here’s the thing about AI coding tools: they work by sending code context to external servers for processing. In English, that means when an engineer at Meta asks Claude Code to help debug a model training script, chunks of that codebase potentially travel outside Meta’s walls.
Anthropic updated its consumer terms in August and September of 2025 to allow opt-in training on select datasets, a revision that likely sharpened the attention of legal and security teams at companies like Meta.
Both Anthropic and OpenAI have been competing aggressively through 2025 and into 2026, adjusting usage limits and pricing strategies. Current plans from both companies sit around the $20 per month range for Pro and Plus tiers. That accessible price point has accelerated adoption inside engineering organizations, which is precisely what makes the data exposure question harder to ignore at scale.
