Breaking

Meta Curbs Internal Use of Claude Code and Codex Over Distillation Concerns

June 29, 2026 at 14:17 EDT

AI Agents
Software Dev & Coding
Sales & Business

Meta is tightening restrictions on employees' use of Anthropic's Claude Code and OpenAI's Codex, it has emerged. The move, in effect as of June 2026, aims to reduce dependence on rival AI coding tools while the company builds its own alternatives. Internal documents indicate Meta has ordered the suspension of certain tasks to avoid legal and technical risks.

June 29, 2026 · Meta

Meta Curbs Use of Claude Code and Codex to Avoid "Contamination"

Internal documents show Meta restricting its applied AI engineers from rival coding tools — fearing that competitors' outputs could leak into its own models through "distillation," while it works to cut dependence on Anthropic and OpenAI.

tools restricted — Claude Code & Codex

$150K+

monthly token spend reported in heavy Claude Code use

firms walling off tools — Meta, Anthropic, Amazon

The policy shift

Before

Prioritize best-in-class tools

→

After

Reduce dependence on outside vendors

Why the restriction

Fear that rival outputs could "distill" into Meta's own models

IP & capability leakage

Competitors' outputs pulled into in-house AI work

Contract / ToS risk

Possible terms-of-service violations

Strategic dependence

Cut reliance while refining the Llama series

Industry-wide pattern

Anthropic moves to a "walled garden," Amazon bans rival tools to boost its Kiro agent — a broad push to protect proprietary AI ecosystems.

Developer pushback

Anthropic's tightening draws criticism for hurting third-party tools like OpenCode and forcing costlier API pricing.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.