Amazon engineers are reaching a breaking point with company-imposed limits on Claude Code, Anthropic’s AI coding assistant that many developers rely on. The restrictions have sparked internal backlash, with engineers arguing the rules disrupt their workflow and slow development.
The irony cuts deep. Amazon has a total investment of $8 billion in Anthropic, yet employees say they can’t freely use Claude Code for production work without approval. This clash reveals a familiar tension for large organizations: how to adopt AI tools quickly without losing control of security, compliance, and risk.
Inside Amazon’s AI tool limits
Despite that massive Anthropic investment, Amazon has tightened rules around where and how engineers can use Claude Code. Internal guidance limits the use of third-party AI coding tools for production code or live products unless teams go through a formal approval process, Business Insider reported.
Amazon has also pushed teams toward its internal coding assistant, Kiro, as the preferred option for production work. That shift has fueled criticism from engineers who argue the policy is too restrictive and doesn’t reflect how widely these tools are already used in day-to-day development.
Amazon publicly champions AI innovation at conferences and on earnings calls. Internally, employees describe a more cautious approach, with guardrails that prioritize risk management even if they limit some of the productivity gains developers expect from AI coding tools.
Backlash from the dev floor
The pushback has been visible on internal forums. In one discussion thread, roughly 1,500 employees supported a call for Amazon to formally adopt Claude Code, according to Business Insider.
Security and compliance concerns sit at the center of Amazon’s stance. The company has to protect proprietary code and customer data. But internal critics argue the current approach feels like a broad restriction rather than a targeted set of controls that match specific risks, creating friction for teams trying to move quickly.
The bigger policy fight
Amazon’s internal dispute is a preview of what more companies will face as AI coding tools shift from experimental to routine. The question isn’t whether engineers will use AI assistance, but what rules will govern how those tools are approved, monitored, and trusted in production environments.
How Amazon responds will be watched across the industry. If a company with Amazon’s scale struggles to balance AI tool access with corporate oversight, smaller organizations may face the same tradeoffs with fewer resources to manage them.
For now, Amazon engineers remain caught between the promise of AI-assisted productivity and the constraints that limit when and where they can use those tools.
Also read: CISA’s new AI security guidance shows what “guardrails” look like when AI moves into critical systems.