Business Insider reports that Amazon employees are not allowed to use Anthropic’s Claude Code in their production work without official approval. Despite other major technology companies encouraging widespread use of AI coding tools, this restriction applies to actual product and production code. Specifically, this is in contrast to software giant Microsoft, which recently asked its own engineers to test Claude Code alongside GitHub Copilot and share feedback on its performance. Last year, Amazon shared an internal guidance policy directing users to use Kiro, its in-house AI coding assistant, according to a BI report. The guidance advises employees to rely on Kiro instead of unapproved third-party tools such as Claude Code when writing production code.
Amazon employee “dissatisfied”
The move sparked criticism across Amazon’s internal discussion forums, the publication reported. In a thread cited in the report, approximately 1,500 employees supported a request to formally approve the Claude code for internal use. Some employees said the restriction created confusion, especially for engineers working on AWS Bedrock, which provides customers with access to third-party AI models, including Claude.“Customers will ask why they should trust or use a tool that they haven’t approved for internal use,” one employee said.Several employees wondered how they could promote Claude Code to customers given its restricted use in official business. One employee wrote internally that customers might wonder why they should trust a tool that Amazon doesn’t allow for use in internal operations.
Amazon employee claims Claude is better than Kilo
This restriction also led to internal debates over productivity. Some engineers claimed that Claude Code was better than Kiro for certain tasks, and warned that limiting tool selection could slow development. Some said the lack of transparency surrounding the decision not to formally approve the Claude Code, especially after early internal documents suggested it had passed security and legal scrutiny, was frustrating for some. That language was later removed, according to the report.
No explicit prohibition against Claude: Amazon spokesperson
An Amazon spokesperson said that Amazon maintains a “strong strategic partnership with Anthropic” but does not specifically ban Claude Code. However, the spokesperson added that the company has “more stringent requirements, particularly for tools used to develop production code,” and that Amazon offers a process for requesting exceptions.He also said the company has seen improvements in efficiency and delivery with Kiro and wants its employees to use internal tools to speed up deliveries to customers. “We are seeing incredible improvements in efficiency and delivery with Kiro, and our customer growth is rapidly accelerating. We want all of our internal employees to be able to take advantage of this capability and deliver to our customers faster,” a spokesperson said as quoted in the report. “While we will continue to support existing tools currently in use, we have no plans to support additional third-party AI development tools.”
