AWS AI Updates: September 1, 2026

1. AWS Agent Registry Reaches General Availability

AWS made AWS Agent Registry generally available, providing a private, governed catalog and discovery layer for agents, tools, skills, MCP servers, and custom resources across an organization. The service offers semantic and keyword search, a console browse experience, and auto-detection that scans AgentCore runtimes and gateways organization-wide to keep records current without manual upkeep. Governance features include approval workflows for resource submissions, tagging for access control, CloudTrail audit trails, and provisioning through CloudFormation, Terraform, and CDK, and the registry is itself exposed as an MCP server for IDE integration alongside AWS Console, CLI, and SDK access. It launched in five regions including US East (N. Virginia), US West (Oregon), Europe (Ireland), Asia Pacific (Tokyo), and Asia Pacific (Sydney), giving teams a way to avoid duplicated agent work and enforce consistent controls at scale. Source

2. Amazon Quick Integrates With AWS Agent Registry

AWS connected Amazon Quick to AWS Agent Registry, letting users search and browse registered MCP servers and agents built with Amazon Bedrock AgentCore and enable them with minimal setup. Connection details are pre-populated automatically from the registry, so business users no longer manually configure links to agents and tools that already exist, and the discovered resources can be used across Quick chat interfaces, agents, applications, flows, and research features. The integration is available in six regions including US East (N. Virginia), US West (Oregon), Europe (Frankfurt), Europe (Ireland), Asia Pacific (Tokyo), and Asia Pacific (Sydney), and is configured through the Quick admin console under Manage Account, Permissions, AWS Agent Registry. It closes the gap between technical teams who build agents and the business users who consume them in Quick. Source

3. Automated Security Response Adds a Generative AI Remediation Toolkit

AWS added an AI Remediation Toolkit to Automated Security Response on AWS, using generative AI to draft custom remediations through any AI assistant with built-in safety guardrails. The toolkit targets the historical requirement that security teams have deep AWS Systems Manager Automation expertise to author remediation logic, using guided prompts and safety mechanisms to compress what took weeks of manual work into hours while reducing configuration errors. It ships alongside expanded coverage for credential compromise, unpatched vulnerabilities, and sensitive data exposure, automatic remediation of Amazon Inspector, GuardDuty, and Macie findings, an enhanced console managing more than 100 security controls, and multi-channel notifications to Email, Slack, Jira, and ServiceNow with severity filtering. The solution deploys across all commercial and opt-in AWS regions, including GovCloud (US) and China regions. Source