Amazon Web Services (AWS) has announced that OpenAI Codex, a powerful AI assistant for code generation and editing, can now operate through Amazon Bedrock. This new integration allows developers to leverage Codex models for various tasks such as understanding unfamiliar code and automating repetitive engineering workflows while maintaining alignment with enterprise AWS identity and access management (IAM) policies.
The Codex model runs locally on developers' workstations but routes inference requests through Amazon Bedrock. Organizations can choose from multiple access patterns—Native AWS Access, AgentCore Gateway, or custom Large Language Model (LLM) Gateway—each offering different levels of control, routing, and policy enforcement. This flexibility caters to different operational and security requirements without duplicating model-access infrastructure.
A recommended approach to adoption begins with a small least-privilege pilot using Native AWS Access, which relies on the company's existing workforce identity providers integrated via AWS IAM Identity Center. This pilot ensures secure authentication, model availability, and telemetry collection before expanding access to wider teams or integrating gateways that provide additional routing or guardrail controls.
Operational observability is a key component of the rollout strategy. AWS provides integration with the AWS Distro for OpenTelemetry and Amazon CloudWatch, allowing organizations to monitor Codex client activity, audit usage patterns, and correlate identity and billing data for governance and cost allocation. This layered telemetry approach improves transparency without conflating token usage with direct cost reports.
To standardize the developer experience and maintain security, organizations are encouraged to deploy approved plugins, hooks, and Model Context Protocol (MCP) servers. These tools help distribute vetted integrations and enforce organizational policies, reducing the risk of unauthorized data exposure or unsafe code changes.
While the integration offers significant benefits, it requires careful configuration of identity roles, region-specific model availability, and telemetry settings. Enterprises must validate that the chosen access path meets security and operational goals and perform rollback testing to ensure safe decommissioning if necessary.
In summary, the integration of OpenAI Codex with Amazon Bedrock provides a secure, scalable path for enterprises to harness AI-assisted coding within AWS while preserving compliance and operational integrity. AWS offers detailed example configurations and validation procedures in the Guidance for Codex on AWS open source repository to aid adoption.
Source & further reading
OpenAI Codex on Amazon Bedrock ↗Source published 2026-11-11. This report was prepared with AI assistance from the linked primary-source material. Claims and limitations should be checked against the original announcement or research.