AWS News - 2026-09-17

Amazon Connect Customer provides a library of standard metrics. Custom metrics let you go further; defining the measurements unique to your business. But managing them across various instances manually means definitions slowly drift until the numbers no longer align. A metric you can't trust is worse than no metric at all, because people still make decisions on it.

Now you define a metric once and manage it programmatically through seven new API operations (CreateMetric, DeleteMetric, DescribeMetric, ListMetrics, SearchMetrics, UpdateMetricContent, and UpdateMetricMetadata) so your definitions stay consistent as you scale across queues, teams, and environments. For example, a customer experience leader noticing a drop in self-service completion rates can define a custom metric that tracks the percentage of interactions where customers abandon and request a live agent within 90 seconds. Using the API, an AI-powered operations assistant can automatically create the metric and publish it to a monitoring dashboard — no manual console work, no definition drift across environments. Every change is tracked in AWS CloudTrail, giving you a complete audit trail.

Amazon Keyspaces (for Apache Cassandra) is now generally available in 11 additional Regions, allowing customers in those Regions to build Cassandra-compatible applications with lower latency while keeping their data within the Region to meet data residency requirements.. 

Amazon Keyspaces (for Apache Cassandra) is a scalable, highly available, and managed Apache Cassandra–compatible database service. Amazon Keyspaces is serverless, so you pay for only the resources that you use and you can build applications that serve thousands of requests per second with virtually unlimited throughput and storage. 

To find the list of new available Regions, see the following resources: AWS Capabilities by Region.

Amazon Elastic Container Service (Amazon ECS) now supports Amazon S3 Files for tasks running on the Amazon EC2 launch type, enabling customers to connect their containerized applications directly to data in Amazon S3 as a shared file system. With this launch, customers running ECS workloads on EC2 instances can work with their S3 data using standard file system semantics, so file-based applications, AI agents, and data processing workloads run on S3 data with no code changes and without duplicating or staging files first. S3 Files support was previously available for ECS tasks on AWS Fargate and ECS Managed Instances, and now extends to the EC2 launch type, giving customers consistent access to S3 Files across all three ECS launch types.

Amazon S3 Files, built on Amazon EFS, delivers a shared file system that connects AWS compute resources directly with data in Amazon S3, providing full file system semantics and low-latency performance without data leaving S3. S3 Files works with new and existing data in S3 buckets, with no migration required. Customers can mount an Amazon S3 Files volume in their Amazon ECS tasks to read and write data in an S3 bucket using standard file system operations, with no application code changes and no need to copy or stage data.

Amazon S3 Files support is available in all AWS commercial Regions and the AWS GovCloud (US) Regions. To learn more, refer our documentation.

Amazon Connect Customer now supports tag-based access control for custom metrics, so you can govern which users and roles can view, create, or modify each custom metric.

As more teams across your organization adopt custom metrics, they end up editing the same shared library. Without controls, one person can change a metric that other teams count on. The usual fix is to give only a handful of people permissions to custom metrics, but that shuts out every other team that wants to start using them.  

Now you can tag each metric, then use those tags to set access: a team gets full control of their own metrics, view-only access to another team’s, and no visibility into the metrics you restrict. Search follows the same access rules, so people only see the metrics they’re allowed to view or manage.

Amazon Connect Customer now lets managers import a PDF of an evaluation form from a third-party quality management application, and automatically recreates it in Connect using AI, making it easier to migrate an existing quality program into Connect. Instead of rebuilding an existing form question by question, you upload the PDF and choose the preferred scoring method, either percentage or points-based, and Connect Customer automatically extracts the sections, questions, answer options, and scoring into a draft evaluation form. You can also provide natural language instructions to import the form more accurately or to adjust it as it's created — for example, splitting a broad question into simpler ones — minimizing the manual edits needed before you activate the form.

This feature is available in the following AWS Regions: US East (N. Virginia), US West (Oregon), Canada (Central), Asia Pacific (Singapore), Asia Pacific (Sydney), Asia Pacific (Tokyo), and Europe (Frankfurt). To learn more, please visit our documentation and our webpage. For information about Amazon Connect Customer pricing, please visit our pricing page.

AWS Client VPN now supports MacOS 27 Golden Gate client with versions 6.0+ You can now run the AWS supplied VPN client on the latest MacOS versions. AWS Client VPN desktop clients are available free of charge, and can be downloaded here.

AWS Client VPN is a managed service that securely connects your remote workforce to AWS or on-premises networks. It supports desktop clients for MacOS, Windows x64, Windows Arm64 and Ubuntu-Linux. With client version 6.0 onwards, Client VPN now supports the latest MacOS  Golden Gate 27.0. It already supports Mac OS version 13.0, 14.0, 15.0, and 26.0, Windows 10 (x64) and Windows 11 (Arm64 and x64), and Ubuntu Linux 22.04 and 24.04 LTS versions.
To learn more about Client VPN:

• Visit the AWS Client VPN product page
• Read the AWS Client VPN documentation
• Read the AWS Client VPN user guide

Amazon Elastic Container Service (Amazon ECS) now provides a consolidated deployment view for Amazon ECS Managed Daemons, giving you a single place to see how far a deployment has progressed, what is failing, and what could stop it. This helps you monitor Managed Daemon deployments at a glance and resolve issues faster, whether a rollout is in progress or you are reviewing what happened after it completed, eliminating the need to piece together deployment status from multiple sources.

The deployment view presents a lifecycle timeline with timestamps at each step and the total deployment duration, including the rollback path if a deployment is interrupted. Progress bars for each capacity provider track instances completed, in progress, and remaining, as well as instances draining and being replaced when a capacity provider is removed. A monitoring panel shows deployment circuit breaker, deployment alarm, and container health check state, each with live Amazon CloudWatch alarm detail and whether it can stop the deployment. If a daemon task fails, the affected capacity provider shows the stop reason with links to the task, its logs, and the matching troubleshooting guide, including the failure that caused a rollback. The view also displays target and source revisions with instance counts, drain percentage, and bake time.

This enhancement is available as a console-only view in all AWS Commercial Regions at no additional cost. To get started with Amazon ECS Managed Daemons, refer to our documentation.

Amazon SageMaker AI now supports serverless model customization for the NVIDIA Nemotron 3.5 Lightning model using supervised fine-tuning (SFT), Direct Preference Optimization (DPO), and reinforcement fine-tuning (RFT). This is one of the latest open-weight models from NVIDIA and employs a hybrid Mixture-of-Experts architecture with 3B active parameters and 30B parameters in total. In addition to deploying this model on SageMaker AI, you can now adapt it to your specific domains and workflows.

Model customization enables you to tailor foundation models with your proprietary data so a smaller, right-sized model can match frontier-model quality on your tasks, reducing cost and latency. You can use labeled data with SFT to improve accuracy on domain-specific tasks, preference data with DPO to align outputs with your organization's tone, or reward signals with RFT to enhance performance on new tasks. With serverless customization, SageMaker AI handles all infrastructure provisioning and training orchestration, so you can focus on your data and evaluation rather than cluster management, and only pay for what you use.

Serverless model customization for NVIDIA Nemotron 3.5 Lightning on SageMaker AI is available in US East (N. Virginia), US West (Oregon), Asia Pacific (Tokyo), and Europe (Ireland). To get started, navigate to the Models page in Amazon SageMaker Studio to launch a customization job, or use the SageMaker Python SDK for programmatic access. To learn more, see the Amazon SageMaker AI model customization documentation. 

Amazon Quick now expands Generate Analysis with two new ways to create dashboards faster. You can generate a single sheet inside an existing analysis by describing it in natural language, and you can generate a new analysis from an image of an existing dashboard.

 

With Generate Sheet, you describe the sheet you want and Amazon Quick adds it to your current analysis with visuals selected for your data, filter controls, and calculated fields such as year-over-year growth and month-over-month comparisons. You can extend an analysis without building each visual by hand.

 

With generate an analysis from an image, you attach an image of a dashboard, including dashboards from other BI tools, to your prompt when you generate an analysis. Amazon Quick recreates it as an editable analysis, building what is supported in Amazon Quick. Both capabilities work with existing publishing workflows, embedding, CI/CD pipelines, and point-and-click editing.

 

At launch, Generate Sheet and generate analysis from an image are available to Enterprise subscription/Author Pro users. Authors also have promotional access to this capability through December 2026 as part of Amazon Quick Enterprise, provided their organization has not restricted access.

 

To learn more, see Generating an analysis with natural language prompts in the Amazon Quick User Guide. To get started, open an analysis and choose Generate Sheet, or attach an image to your prompt and choose Generate analysis.

Amazon Simple Email Service (SES) now supports tenant-level deliverability insights in Virtual Deliverability Manager (VDM). Tenant management allows email senders to isolate email sending across customers, business units, or applications. Previously, VDM provided deliverability metrics at the account, ISP, sending identity, and configuration set levels. Now, customers who use tenants and have enabled VDM can also monitor the deliverability of each tenant.

The VDM dashboard adds a Tenants view with per-tenant metrics (send volume, deliveries, bounces, complaints, opens, and clicks) and a detail page that breaks those metrics down by mailbox provider and by associated identities and configuration sets. Customers can also search and export a tenant's sent messages, or query tenant metrics programmatically using the BatchGetMetricData API operation's new TENANT_NAME dimension. This helps senders identify the tenant causing a deliverability issue and correct it without impact to their other tenants.

Virtual Deliverability Manager is available in all AWS Regions where Amazon SES is available.

To learn more, visit the Amazon SES console or refer to the Virtual Deliverability Manager dashboard documentation in the Amazon SES Developer Guide.

AWS Builder Center is a platform that brings together the global builder community, offering articles, workshops, sandbox environments, events, and learning resources in one place. Previously limited to desktop and web access, the experience is now available as a mobile app on iOS and Android. Builders worldwide can stay connected and engaged with the community from anywhere, using their AWS Builder ID to remain signed in across sessions.

The AWS Builder Center mobile app enables builders to browse trending articles, access 600+ AWS Skill Builder courses, and manage hands-on workshops with free sandbox environments directly from their mobile devices. Builders can follow AWS Heroes, Community Builders, and User Group Leaders, check Builder Loft event calendars on the go, and receive push notifications for subscribed topics and communities. The app also supports the Wishlist feature, allowing builders to submit and manage product feedback directly to AWS teams.

The AWS Builder Center mobile app is available worldwide on the Apple App Store and Google Play Store.

To learn more about the AWS Builder Center mobile app, visit the AWS Builder Center mobile page.

The Amazon S3 Express One Zone storage class is now available in 7 additional AWS Regions: Asia Pacific (Singapore), South America (Sao Paulo), US West (N. California), Canada (Central), Europe (Paris), Asia Pacific (Sydney), and Asia Pacific (Seoul).

Amazon S3 Express One Zone is a high-performance, single-Availability Zone storage class purpose-built to deliver consistent single-digit millisecond data access for your most frequently accessed data and latency-sensitive applications. S3 Express One Zone delivers data access speed up to 10x faster and request costs up to 80% lower than S3 Standard. It enables workloads such as machine learning training, interactive analytics, and key-value caching in AI search engines to achieve fast data access with high durability and availability.

With this expansion, S3 Express One Zone is now available in 15 AWS Regions. For pricing details, visit the S3 pricing page. To learn more, visit the product page and documentation.

Revised slightly: AWS has reimagined the getting started experience with smart and sensible defaults to help developers get started fast so that they can focus on building. New customers can sign up and get started right away with $100 in Free Tier credits, managed project environments, and simplified team and resource permissions. If their workload evolves, customers can activate advanced features to access the full breadth and depth of AWS in one step, with no migration required.

Kiro の IDE、CLI、Web のすべてで、GPT‑5.6 Sol、Terra、Luna が 100 万トークンのコンテキストウィンドウに対応します。これにより、コードベース全体、長文ドキュメント、長く続いたマルチターンのエージェント履歴を、チャンク分割や要約を挟まず、細部を失わないまま 1 回のリクエストで処理できるようになります。

技術的負債への従来のアプローチは、開発者の時間を保守に割り当てて機能提供を遅らせるか、負債が複利で膨らむのを放置するかの二択を強います。対話型の AI コーディングアシスタントは個々のタスクを速くしてくれますが、それでもすべてのステップで開発者がループの中にいる必要があります。私たちが欲しかったのは第三の選択肢、つまり issue を読むところからプルリクエストを開くところまで保守タスクを一気通貫で引き受け、エンジニアは判断が最も価値を持つコードレビューでだけ関わる、という形です。それが Web の Kiro の autonomy モードです。保守を含む幅広いタスクをこなせるエージェンティックなソフトウェアエンジニアリングツールで、エンジニアはより価値の高い仕事に集中できます。実際にどうなったかというと、2 か月のあいだに Kiro は私たちのリポジトリ群で 87 件のオープン issue に対応するプルリクエストを提出しました。これは、それまでの 22 か月（約 2 年）でチームが対応してきた量に迫る数です。本記事では、どのようにセットアップしたか、2 つのケーススタディの詳細、そして自分のリポジトリで Kiro を自律的に使う方法を紹介します

店舗業務支援 AI エージェント SMART を起点に、ファッション・アパレル業界 6 社 20 名が集まった第一回ナレッジ共有会の開催報告です。週報作成 15 分が 0 分に、IT 未経験の店舗出身メンバーが 2 週間で Web アプリ開発、ベテランの経験に依存していた在庫消化業務のエージェント化——SMART の合同ブートキャンプから活動を始めた 3 社が、実際に手を動かし、現場との対話を重ねて得た実践ナレッジを発表しました。作って終わりにしない AI エージェント開発のヒントが詰まった当日の模様をレポートします。

Rust の型システムは多くのバグを防ぎますが、コードが本当に正しいことまでは保証しません。そこで、Rust 向けのオープンソースの自動プログラム検証ツール Verus を紹介します。事前条件・事後条件による仕様記述の書き方、1 秒未満の高速なフィードバック、unsafe コードや並行コードの正しさの証明、そして Nitro Isolation Engine をはじめとする Amazon 社内やオープンソースでの活用事例まで説明します。

The AWS European Sovereign Cloud is a new, independent cloud for Europe, physically and logically separate from existing AWS Regions and operated within the European Union (EU). It provides the same services, features, and APIs as AWS commercial Regions, but runs as a distinct AWS partition (aws-eusc), with its own control plane, AWS Identity and […]

Bulletin ID: 2026-113-AWS
Scope: AWS
Content Type: Important (requires attention)
Publication Date: 09/16/2026 12:30 PM PDT

Description:

Network Policy Agent is an EKS Policy management feature. We identified CVE-2026-86831, a cross-namespace NetworkPolicy bypass in Amazon EKS Network Policy Agent (aws-network-policy-agent) before v1.4.0. Pod identifiers are constructed by concatenating the pod name and namespace with a hyphen delimiter, which is a legal character in both Kubernetes pod names and namespace names. This ambiguity can produce identical identifiers for pods across different namespaces, potentially allowing NetworkPolicy enforcement to be bypassed.

Impacted versions:
- 1.14.0<1.22.3 Amazon VPC CNI Managed Add-on
- < v1.4.0 Network Policy Agent

Please refer to the article below for the most up-to-date and complete information related to this AWS Security Bulletin.

Learn how to automate end-to-end PII detection and redaction from scanned documents at scale using Amazon Bedrock Data Automation with a custom blueprint, AWS Step Functions, and AWS Lambda. A custom blueprint redacts sensitive fields with field-level precision, and a token matching quality check raises recall across degraded and handwritten documents.

AgentCore optimization turns production traces into proposed configuration changes, then validates them before promotion. This technical companion to the launch post explains how the system prompt optimizer's reflector engine works and shares benchmark results for the Single Agent and Sub-Agent Reflectors.

Integrate NVIDIA Resiliency Extension (NVRx) into PyTorch FSDP training on Amazon EKS to overlap checkpoint I/O with training and recover from GPU faults in seconds. This post covers async checkpointing, in-process restart, and ft_launcher in-job restart, with H100 benchmarks at 2 to 8 nodes showing 99%+ training efficiency and second-scale recovery.

AI agents on foundation models often misapply healthcare and life sciences decision frameworks, citing the right guideline but applying it incorrectly. This post shares 38 open-source agent skills across 11 HCLS domains that close this gap, with installation steps, three worked use cases, and a 410-prompt evaluation showing a 70-86% win rate.