AWS News - 2026-09-05

Amazon GuardDuty now offers Custom Detection Rules, a library of 35 prebuilt, opt-in rules for CloudTrail management events that let you extend threat detection coverage to match your environment. Custom Detection Rules produces 26 unique finding types mapped to 10 MITRE ATT&CK® tactics, without the heavy lifting of log ingestion, normalization, or storage.

Some threat techniques, such as sharing an AMI externally, disabling flow logs, or signing in without MFA, could be meaningful indicators of compromise in some accounts but routine in others. Custom Detection Rules lets you enable these detections only where the activity is unexpected — expanding your TTP coverage tailored to your environment.

To get started, browse Custom Detection Rules via the GuardDuty console or API, and enable rules in dry-run mode to evaluate detection efficacy before going live.

Custom Detection Rules is available in all AWS commercial Regions and the AWS GovCloud (US) Regions. To learn more, see Amazon GuardDuty Custom Detection Rules. To receive programmatic updates on new Amazon GuardDuty features and threat detections, subscribe to the Amazon GuardDuty SNS topic.

Amazon EC2 now enables AMI owners to define which instance types are compatible with their AMIs. Owners can specify supported instance types, unsupported instance types, or both — and any launch attempt on a non-permitted instance type is automatically blocked.

AMI owners now have a built-in way to prevent launches on instances that are not compatible with their AMIs. This reduces the risk of failed launches due to incompatible instance-AMI pairings. By default, an AMI can be launched on any instance type, so existing workflows remain unaffected until restrictions are explicitly applied.

This feature is available in all AWS Regions at no additional cost. To learn more, please visit the documentation.

AWS announces automatic sync scheduling for Amazon Bedrock Managed Knowledge Base, a fully managed retrieval-augmented generation (RAG) service that handles data ingestion, storage optimization, and advanced retrieval without requiring you to manage vector databases or data pipelines. Previously, keeping your knowledge base current required manually triggering a sync each time your source data changed or building a custom solution. Now, you can configure daily, weekly, or monthly sync schedules for all native data source connectors, so your AI agents always retrieve the most up-to-date information.
With automatic sync scheduling, you can match your sync frequency to how often your source content changes. For example, set a daily sync for a rapidly evolving customer support knowledge base in Confluence, a weekly sync for SharePoint policy documents that update periodically, or a monthly sync for reference materials stored in Amazon S3. This eliminates the need to build and maintain custom scheduling workarounds, reducing operational overhead while ensuring your retrieval-augmented generation applications stay grounded in current enterprise data.
To learn more, see Sync scheduling for data sources in the Amazon Bedrock User Guide. For more information about Amazon Bedrock Managed Knowledge Base, visit the Amazon Bedrock Knowledge Bases product page.

AWS announces the ServiceNow data source connector for Amazon Bedrock Managed Knowledge Base, a fully managed retrieval-augmented generation (RAG) service. Customers can now connect their ServiceNow instance to crawl knowledge articles and service catalog items directly into their managed knowledge base. Previously, bringing ServiceNow content into Bedrock Knowledge Bases required building and maintaining custom ingestion pipelines—now, you simply provide your ServiceNow instance credentials, and the connector handles data crawling, metadata extraction, and incremental sync automatically.
The ServiceNow connector crawls knowledge articles and service catalog items, including file attachments, giving your AI agents access to the institutional knowledge already maintained in ServiceNow. You can scope crawls to specific knowledge bases, article categories, or service catalogs using sys ID inclusion lists, ensuring only relevant content is ingested. This makes it straightforward to power employee-facing IT assistants, HR helpdesks, or customer support agents grounded in your organization's up-to-date ServiceNow content.
To learn more, see ServiceNow data source in the Amazon Bedrock User Guide. For more information about Amazon Bedrock Managed Knowledge Base, visit the Amazon Bedrock Knowledge Bases product page.

AWS announces user-managed setup (3LO) for SharePoint, OneDrive, and Confluence data sources in Amazon Bedrock Managed Knowledge Base. Previously, configuring these data sources required generating 2LO credentials on the third-party side, which could be time-consuming and inaccessible for users who lack admin-level access to those systems. With user-managed setup, you simply sign in with your existing third-party credentials, and Amazon Bedrock Managed Knowledge Base handles authentication—allowing you to complete data source setup in just a few minutes.
This streamlined experience lowers the barrier to getting started with managed knowledge bases. Teams that want to quickly prototype an AI assistant grounded in their SharePoint documentation, OneDrive files, or Confluence wikis no longer need to coordinate with IT administrators to obtain service account credentials. User-managed setup complements the existing service account authentication, giving you a faster path to connect your data while still preserving the programmatic, enterprise-grade option for production workloads.
To get started, see the following resources:; SharePoint user-managed setup in the Amazon Bedrock User Guide, OneDrive user-managed setup in the Amazon Bedrock User Guide and, Confluence user-managed setup in the Amazon Bedrock User Guide

Amazon Managed Workflows for Apache Airflow (MWAA) Serverless is now available in AWS GovCloud (US-East) and AWS GovCloud (US-West). Amazon MWAA Serverless eliminates the operational overhead of managing Apache Airflow infrastructure by automatically provisioning and scaling compute resources on demand, with a pay-per-use model that charges only for actual workflow run time.

For a full list of supported regions, see the Amazon MWAA Serverless regions page. To learn more, see the Amazon MWAA Serverless documentation.

We’re pleased to confirm the successful completion of our annual Amazon Web Services (AWS) Outsourced Service Provider’s Audit Report (OSPAR) assessment on July 29, 2026, in line with the OSPAR version 2.0 framework. The Association of Banks in Singapore (ABS) established the Guidelines on Control Objectives and Procedures for Outsourced Service Providers (ABS Guidelines) to […]

Bulletin ID: 2026-098-AWS
Scope: AWS
Content Type: Important (requires attention)
Publication Date: 09/04/2026 10:30 AM PDT

Description:

log4j-cve-2021-44228-hotpatch is a tool which injects a Java agent into a running JVM process. The agent will attempt to patch the lookup() method of all loaded org.apache.logging.log4j.core.lookup.JndiLookup instances to unconditionally return the string "Patched JndiLookup::lookup()". It is designed to address the CVE-2021-44228 remote code execution issue in Log4j without restarting the Java process. We identified CVE-2026-85656, an OS command injection issue in the log4j-cve-2021-44228-hotpatch package in Amazon Linux before 1.3-9.amzn2 might allow a local user to execute arbitrary commands with root privileges via a Java process whose executable path contains embedded newline characters.

Impacted versions: <=1.3-8.amzn2

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

Bulletin ID: 2026-099-AWS
Scope: AWS
Content Type: Important (requires attention)
Publication Date: 09/04/2026 11:45 AM PDT

Description:

The Amazon EFS CSI Driver is an open-source Kubernetes Container Storage Interface (CSI) driver that lets Kubernetes workloads use Amazon EFS file systems. We identified CVE-2026-85781, an issue in the driver's volume-deletion logic. When the controller is configured with the non-default --delete-access-point-root-dir=true option, it did not verify that the EFS access point referenced by a PersistentVolume's volume handle belonged to the file system referenced by that same volume handle. An authenticated Kubernetes user with permission to create PersistentVolumes could use a crafted volume handle to cause the controller to recursively delete directories from an Amazon EFS file system that the user was not otherwise authorized to modify. This issue does not affect Amazon EFS itself; Amazon EFS access controls operated as designed.

Impacted versions: <=3.4.0

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

Bulletin ID: 2026-100-AWS
Scope: AWS
Content Type: Important (requires attention)
Publication Date: 09/04/2026 12:30 PM PDT

Description:

ion-java is a Java library that implements the Amazon Ion data format specification. We identified CVE-2026-85786, memory-amplification denial of service via highly compressed data expansion. ion-java 1.12.0 added a GZIP auto-decompression opt-out for CVE-2026-75936, but the implementation of the opt-out in 1.12.0 was insufficient to address the issue.

Impacted versions: < 1.12.1

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

Bulletin ID: 2026-097-AWS
Scope: AWS
Content Type: Important (requires attention)
Publication Date: 09/04/2026 10:00 AM PDT

Description:

Amazon awslabs.dynamodb-mcp-server is an open-source Model Context Protocol (MCP) server that enables AI coding assistants to interact with Amazon DynamoDB, including table design, data modeling, and CDK infrastructure generation. We identified CVE-2026-85654, an improper neutralization of special elements used in a template engine in the CDK generator component. Under certain circumstances, a context-dependent actor could execute arbitrary code on the host that deploys the generated application via crafted table, index, or attribute names in a data model file.

Impacted versions: >= 2.0.10 AND <= 2.1.5

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

Bulletin ID: 2026-101-AWS
Scope: AWS
Content Type: Important (requires attention)
Publication Date: 09/04/2026 13:00 PM PDT

Description:

We have identified CVE-2026-85787, an incomplete list of disallowed inputs in the SQL validation component in Amazon awslabs postgres-mcp-server before version 1.1.7 might allow an unauthenticated actor to modify data beyond the read-only scope by placing crafted SQL into the content that is submitted when an authenticated user interacts with the MCP server.

Impacted versions: any pypi package version < 1.1.7

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

Disaster recovery at scale is hard. Learn how Intuit built EWOK Agent, an agentic disaster recovery assistant on Amazon Bedrock that lets on-call engineers run production failovers from a plain-language request while keeping every action audited, policy-compliant, and safe.

Learn how to customize an Amazon Bedrock knowledge base for large, complex documents by combining the high-accuracy text extraction of Amazon Textract with the generative AI of Amazon Bedrock. This post shows how to ingest and preprocess PDFs and images, then query utility bills at scale for faster, more accurate customer interactions.

HyperPod InstantStart is an open source control plane that composes Amazon EKS orchestration with the managed capabilities of Amazon SageMaker HyperPod. It drives the same guarded operations through both a web interface and an AI agent, turning cluster bootstrap, capacity, training, inference, and storage into dependable, agent-driven infrastructure.

Building a Physical AI system takes a continuous pipeline, not a single training job. This post shows how to run that model factory (synthetic data generation, post-training, and closed-loop evaluation with NVIDIA Cosmos 3) on a persistent, resilient Amazon SageMaker HyperPod cluster on Amazon EKS, with GPU goodput as the metric that matters.

Long-running AI agents accumulate outdated memories that degrade quality and create compliance risk. Learn how to design memory lifecycle policies for Amazon Bedrock AgentCore: scoring, consolidating, and pruning agent memories on a nightly AWS Step Functions workflow, with a deployable AWS CDK stack.

Learn how to deploy a multimodal WhatsApp ordering assistant that takes customer orders through text, voice notes, and real-time voice calls on a single business number, built on Amazon Bedrock AgentCore with Amazon Nova 2. The channel and ordering layers stay separate, and one shared memory recognizes each customer across all three channels.