May 2023

Starting today, you can enable a new Managed Domain List on Amazon Route 53 Resolver DNS Firewall, to block domains identified as low-reputation or that are known or suspected to be malicious by Amazon GuardDuty’s threat intelligence. This means that customers using GuardDuty can now block domains using the same GuardDuty threat intelligence used to monitor and alert you on potential DNS threats for your AWS accounts today.

Today, AWS announced the opening of a new AWS Direct Connect location within the Sonda Quilicura Q2 data center in Santiago, Chile. By connecting your network to AWS at the new location, you gain private, direct access to all public AWS Regions (except those in China), AWS GovCloud Regions, and AWS Local Zones.

Starting today, Amazon SageMaker JumpStart provides the capability to fine-tune a large language model, particularly a text generation model on a domain-specific data set. Customers can now fine-tune models with their custom data set to improve performance in specific domains. For example, this blog describes how to use domain adaption to fine tune a GPT-J 6B model on publicly available financial data from the Security and Exchange Commission so that the model can generate more relevant text for financial services use cases. Customers can fine-tune Foundation Models such as GPT-J 6B and GPT-J 6B FP16 models for domain adaptation on JumpStart inside Amazon SageMaker Studio through UI, and through SageMaker Python SDK.

AWS Clean Rooms is now a Health Insurance Portability and Accountability Act of 1996 (HIPAA) eligible service, enabling covered entities and their business associates to use AWS Clean Rooms to process, maintain, and store protected health information. If you have an executed Business Associate Addendum (BAA) with AWS, you can now use AWS Clean Rooms for workloads that are subject to HIPAA compliance.

Amazon Kinesis Data Firehose customers can now send data to Amazon OpenSearch Service using OpenSearch Service auto-generated document ID option. This configuration option enables write-heavy operations, such as log analytics and observability, to consume fewer CPU resources at the OpenSearch domain, resulting in improved performance.

Today, we’re excited to announce that data scientists can use CodeWhisperer for no additional charge to generate real-time code suggestions for Python notebooks in JupyterLab and Amazon SageMaker Studio. With CodeWhisperer, you can write a comment in natural language that outlines a specific coding task in English and CodeWhisperer will recommend one or more code snippets directly in the Notebook that can accomplish the task.

Today, AWS open-sourced the Cedar policy language and authorization engine. You can use Cedar to express fine-grained permissions as easy-to-understand policies enforced in your applications, and you can decouple access control from your application logic. Cedar supports common authorization models such as role-based access control and attribute-based access control. It follows a new verification-guided development process to give you high assurance of Cedar’s correctness and security: AWS formally models Cedar's authorization engine and other tools, proves safety and correctness properties about them using automated reasoning, and rigorously tests that the model matches the Rust implementation.

Amazon Kinesis Data Analytics makes it easier to transform and analyze streaming data in real time with Apache Flink. Apache Flink is an open source framework and engine for processing data streams. Amazon Kinesis Data Analytics reduces the complexity of building and managing Apache Flink applications. Amazon Kinesis Data Analytics for Apache Flink integrates with Amazon Managed Streaming for Apache Kafka (Amazon MSK), Amazon Kinesis Data Streams, Amazon OpenSearch Service, Amazon DynamoDB streams, Amazon Simple Storage Service (Amazon S3), custom integrations, and more using built-in connectors. You can learn more about Amazon Kinesis Data Analytics for Apache Flink here.

We are excited to announce the availability of ml.inf2 and ml.trn1 family of instances on Amazon SageMaker for deploying machine learning (ML) models for Real-time and Asynchronous inference. You can use these instances on SageMaker to achieve high performance at a low cost for generative artificial intelligence (AI) models, including large language models (LLMs) and vision transformers. In addition, you can use SageMaker Inference Recommender to help you run load tests and evaluate the price-performance benefits of deploying your model on these instances.

Amazon Relational Database Service (RDS) for PostgreSQL now supports the pgvector extension to store embeddings from machine learning (ML) models in your database and to perform efficient similarity searches. Embeddings are numerical representations (vectors) created from generative AI that capture the semantic meaning of text input into a large language model (LLM). pgvector can store and search embeddings from Amazon Bedrock, Amazon SageMaker, and more.