Amazon discourages using Anthropic’s Claude Code for production work in favor of its in-house Kiro coding assistant, with formal approvals required for any exceptions.

Internal debate has flared up, including a thread in which roughly 1,500 employees urged formally approving Claude Code for internal use.

Guidance from late 2022 pushes teams to favor Kiro over Claude Code, fueling significant discussion and continued advocacy for Claude Code among employees.

Critics warn about reliability, security, and long‑term effects on engineering practices when adopting AI-assisted coding tools.

Story published February 12, 2026, with Business Insider as the cited source.

Executives and engineers question the lack of transparency around the decision not to formally adopt Claude Code and how data informed the policy.

The policy raises questions for customers about trusting a tool not approved for internal production use and touches on ongoing discussions about its impact on AWS Bedrock integration.

Microsoft is prompting engineers to test Claude Code alongside GitHub Copilot and provide performance feedback.

The dispute reflects broader industry tensions between in-house AI tools and external solutions like Claude Code, with implications for retention and productivity.

Anthropic declined to comment on the matter.

Bedrock sellers face awkwardness promoting Claude Code to customers while it is not allowed for internal production work.

Engineers on AWS Bedrock must navigate a dilemma: cannot officially use Claude Code in development yet promote a platform that supports third‑party AI, risking customer skepticism and weaker demos.