# Securing Cloud Innovation: AWS Updates Across EC2, Security Hub, and AI, with Terraform Integration Status

- Cloud: AWS
- Date: 2026-10-10
- Canonical page: https://jonathanglass.cloud/whats-new/aws/2026-10-10.html
- Dataset: https://jonathanglass.cloud/data/aws/daily_summaries.json

As a Cloud Security Architect and DevSecOps Lead, I'm pleased to present today's AWS feature updates, focusing on their security implications and Terraform integration status.

### Compute Highlights
*   **Amazon EC2 R8g Instances in AWS European Sovereign Cloud (Germany)**: AWS has launched R8g instances in the European Sovereign Cloud, powered by Graviton4 processors. These instances offer up to 30% better performance than Graviton3, with increased vCPU (up to 48xlarge) and memory (up to 1.5TB), making them ideal for memory-intensive workloads like databases and real-time analytics. The availability in the European Sovereign Cloud underscores a commitment to data residency and sovereignty, crucial for highly regulated industries. Architecturally, migrating to Graviton4 requires careful planning for ARM64 compatibility but yields significant performance and cost benefits. Security-wise, Graviton processors offer enhanced hardware-level security features. *Terraform Support*: Supported for instance provisioning via the `instance_type` argument in the `aws_instance` resource in [Terraform AWS Provider: aws_instance](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/instance) in **v5.26.0+**.
*   **Amazon EC2 R8gd Instances in AWS European Sovereign Cloud (Germany)**: Complementing the R8g instances, R8gd instances are now available in the AWS European Sovereign Cloud, featuring up to 11.4 TB of local NVMe-based SSD storage. Also powered by Graviton4, they deliver up to 30% better performance and are built on the AWS Nitro System, making them excellent for applications requiring high-speed, low-latency local storage. A new capability allows customers to adjust network and EBS bandwidth by 25% using EC2 instance bandwidth weighting configuration, offering greater resource optimization. Security considerations include ensuring data at rest on local NVMe is adequately protected (though Nitro-based instances provide hardware-level encryption) and managing access to these high-performance resources. *Terraform Support*: Supported for instance provisioning via the `instance_type` argument in the `aws_instance` resource in [Terraform AWS Provider: aws_instance](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/instance) in **v5.72.0+**. Direct Terraform support for the EC2 instance bandwidth weighting configuration is pending specific provider argument release.

### Security, Identity, & Compliance Highlights
*   **AWS Security Hub Exports to Amazon S3 (CSV or JSON OCSF)**: Security Hub now enables on-demand export of findings to an S3 bucket in either CSV or JSON (OCSF) format. This feature significantly enhances the ability for security teams to integrate Security Hub findings with external SIEM/SOAR systems, data lakes, or for offline analysis and reporting. The OCSF format promotes interoperability across security tools. From a security perspective, it's critical to ensure the target S3 bucket is configured with strong encryption (KMS CMKs), strict access controls (IAM policies, bucket policies), and optionally VPC endpoints for private access, to protect sensitive security findings. This also aids in compliance reporting and auditing. *Terraform Support*: While the underlying S3 bucket and KMS key for findings can be managed using `aws_s3_bucket` and `aws_kms_key` respectively, the direct configuration of the on-demand export feature within Security Hub is pending specific provider resource/argument support. The `aws_securityhub_account` resource manages the service enablement in [Terraform AWS Provider: aws_securityhub_account](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/securityhub_account) in **v5.70.0+**.

### Developer Tools Highlights
*   **New Anthropic Claude 5.5 Models in Kiro for AWS GovCloud (US) Regions**: Kiro IDE and CLI in AWS GovCloud (US) now support Anthropic's Claude Opus 5.5 and Sonnet 5.5 models. Opus 5.5 offers enhanced agentic coding capabilities and adaptive thinking, while Sonnet 5.5 provides faster output and improved performance for coding and knowledge work. Availability in GovCloud is paramount for government agencies and contractors, ensuring compliance with stringent regulatory requirements like FedRAMP. For DevSecOps, leveraging these advanced AI models within a secure GovCloud environment can accelerate secure code development, analysis, and vulnerability detection, provided proper guardrails and human oversight are in place. *Terraform Support*: Pending provider release (use AWS CLI/SDK or custom provisioner in the interim).

### Machine Learning Highlights
*   **Amazon Bedrock Reasoning Summaries for OpenAI Models**: Amazon Bedrock now supports the `reasoning.summary` parameter for OpenAI models via the Responses API. This provides a human-readable summary of the model's reasoning process alongside its response, offering valuable context for evaluating, debugging, and understanding complex AI outputs in tasks like coding or multi-step problem-solving. This feature significantly enhances the explainability and interpretability of AI models, which is crucial for responsible AI development, identifying potential biases, and ensuring generated content aligns with security and ethical guidelines. *Terraform Support*: This feature is an API parameter for OpenAI models within Bedrock and is not directly managed by Terraform resources. However, the Amazon Bedrock service itself and related configurations like guardrails are supported via resources such as `aws_bedrock_guardrail` in [Terraform AWS Provider: aws_bedrock_guardrail](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/bedrock_guardrail) in **v5.60.0+**.

### Contact Center Highlights
*   **Amazon Connect Automated Checks for Performance Evaluation Forms**: Amazon Connect now offers automated suggestions to improve the accuracy of performance evaluation forms, particularly those automatically filled by AI. Managers can run a check to identify questions lacking context for reliable AI scoring and receive suggestions for more specific definitions. This feature is vital for ensuring fair, accurate, and consistent evaluations of both human and AI agents, reducing bias and improving the quality of customer interactions. From a compliance standpoint, consistent evaluation criteria are essential for auditability and regulatory adherence. *Terraform Support*: While Amazon Connect evaluation forms can be managed via the `aws_connect_evaluation_form` resource in [Terraform AWS Provider: aws_connect_evaluation_form](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/connect_evaluation_form) in **v5.72.0+**, the automated check functionality itself is a console/service feature and not a direct Terraform resource configuration.

### Storage Highlights
*   **Amazon S3 Vectors Metadata Pre-filtering in GovCloud Regions**: Amazon S3 Vectors now supports metadata pre-filtering in AWS GovCloud (US-East and US-West) Regions. This capability evaluates metadata filters before running similarity search, returning up to 5x more matching vectors when filters are selective. Additionally, a new `$startsWith` operator has been added for filtering on values like paths and URLs. These improvements enhance the accuracy and completeness of results for Retrieval-Augmented Generation (RAG), agentic, and semantic-search applications. Availability in GovCloud ensures that sensitive government data can leverage these advanced vector search capabilities while adhering to strict compliance requirements. Security considerations include robust access controls on S3 buckets and vector indexes, and proper encryption of both data and metadata. *Terraform Support*: Supported for managing S3 buckets via `aws_s3_bucket` and for configuring S3 Object Vector Indexes via `aws_s3_bucket_object_vector_index` in [Terraform AWS Provider: aws_s3_bucket_object_vector_index](https://registry.terraform.io/providers/hashicorp/aws/latest/docs/resources/s3_bucket_object_vector_index) in **v5.45.0+**. The metadata pre-filtering behavior is inherent to the index configuration.

## Announcements in this summary

- [Amazon EC2](https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-ec2-r8g-instances-thf/) (Compute)
- [Security Hub](https://aws.amazon.com/about-aws/whats-new/2026/10/security-hub-exports-s3-csv-json/) (Security, Identity, & Compliance)
- [Kiro](https://aws.amazon.com/about-aws/whats-new/2026/06/kiro-claude-5-5-aws-govcloud-us/) (Developer Tools)
- [Amazon Bedrock](https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-bedrock-reasoning-summaries-openai/) (Machine Learning)
- [Amazon EC2](https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-ec2-r8gd-thf/) (Compute)
- [Amazon Connect](https://aws.amazon.com/about-aws/whats-new/2026/10/amazon-connect-customer-automated-checks-evaluation-forms/) (Contact Center)
- [Amazon S3 Vectors](https://aws.amazon.com/about-aws/whats-new/2026/10/s3-vectors-metadata-pre-filtering-in-govcloud-regions/) (Storage)
