Version history

AWS Bedrock's service terms were updated in an update detected on July 30, 2026, with several operational refinements. The terms expanded the definition of 'AI Services' to explicitly include AWS IoT SiteWise Scenario Discovery, clarified constraints on its use by adding language prohibiting reliance on its data for vehicle safety determinations, and refined IoT service liability language. These changes establish explicit guardrails around automated decision-making in safety-critical contexts like vehicle systems.

AWS updated its Service Terms on July 18, 2026 with several operational clarifications and restrictions. For On-Demand Capacity Reservations, the terms now explicitly prohibit resale and reserve the right to cancel or terminate instances if resale is suspected. For Capacity Blocks for ML, the termination window was extended from 30 minutes to 60 minutes for UltraServer instance types. Additionally, AWS renamed its Amazon Sidewalk qualification program from 'Works with Amazon Sidewalk' to 'On Amazon Sidewalk' and simplified related trademark and badge usage requirements.

AWS Bedrock updated its AWS Clean Rooms service terms on July 11, 2026, revising how content deletion and resource removal work. The previous language stated that deletion requests would result in removal of all content from the collaboration dataset. The updated language now requires that deletion, removal of configured resources, or disallowing content from being used in intermediate tables will result in termination of access to that content or deletion from intermediate tables, with specifics described in service documentation. This shifts the deletion mechanism from a blanket dataset removal to a more granular resource-level operation tied to service documentation.

AWS Bedrock updated its AWS Service Terms on July 1, 2026, removing several AWS IQ marketplace provisions and adding new language around AWS Config third-party recorders. The change eliminates the 24-month exclusivity requirement for using AWS IQ to pay providers, removes obligations to revoke provider access immediately upon violations, and removes a blanket release of claims related to AWS IQ. The update adds explicit authorization for AWS to use and store Config Content to develop and improve AWS Config and related security services, with an opt-out mechanism available through AWS Organizations.

AWS Bedrock added a new section (110) to its Service Terms governing free exploration services, which are technical consulting and advisory services AWS may provide at no charge. The updated terms establish that AWS may discontinue these services without notice, that customers bear all responsibility for testing and deploying any content or code AWS provides, and that AWS retains the right to develop competing products based on that content. The terms also restrict what types of data customers can share with AWS during these engagements and clarify that customers alone are responsible for ensuring any AI solutions provided comply with applicable law and are production-ready.

AWS Bedrock updated its Service Terms on June 23, 2026 to add Amazon GuardDuty to the list of Services that may incorporate generative AI features powered by Amazon Bedrock. The previous version listed 27 services; the updated version now lists 28. This means Amazon GuardDuty users may now encounter Bedrock-powered AI features in that service, and Amazon Bedrock abuse detection mechanisms may apply to it.

AWS updated its Service Terms on June 18, 2026 to rename a service and clarify which AI-powered services incorporate Amazon Bedrock functionality. The company renamed AWS Security Agent to AWS Continuum and added clarification that Continuum includes features previously named AWS Security Agent. This affects the list of services where Bedrock abuse detection mechanisms apply and the definition of AI Services covered under the terms.

AWS updated its Professional Services licensing language on June 17, 2026. The prior version stated that Developed Content deliverables 'may include' AWS Content or Third Party Content provided under separate licenses. The updated version revises this to say that 'in addition to Developed Content, deliverables may include' AWS Content or Third Party Content. This clarifies the relationship between Developed Content and other content types in Professional Services deliverables, specifying that AWS Content and Third Party Content are distinct from and in addition to Developed Content rather than potentially overlapping categories.

AWS Bedrock updated its service terms on June 16, 2026 to add new provisions governing metadata sharing with Anthropic for certain AI products and to introduce a new AWS WAF AI traffic monetization feature. The updated terms now explicitly authorize AWS to notify Anthropic of metadata present in requests to Anthropic models, and establish a framework whereby AWS can facilitate payment transactions between users and buyers of content served through AWS services, though AWS states it does not provide regulated financial services and is not a party to the flow of funds. Users who employ these features will have pricing, payment, and configuration information shared with applicable buyers, payment providers, and facilitators under separate terms between the user and those parties.

AWS updated its Service Terms on June 10, 2026 to add AWS FinOps Agent (Preview) to the list of AI Services and services that incorporate generative AI features. More substantially, AWS restructured and expanded its abuse detection provisions: it now requires explicit consent from users of certain Anthropic models for AWS to transfer user content and metadata to Anthropic for abuse detection processing, and it clarified that Bedrock stores inputs and outputs for up to 30 days solely for abuse detection purposes on identified models. This shifts abuse detection from an implicit AWS-only process to one that may involve third-party model providers with user consent.

AWS Bedrock updated its abuse detection and monitoring practices on June 2, 2026. The revised terms clarify that abuse detection mechanisms may now store service inputs and outputs for up to 30 days to detect policy violations, and specify that if child sexual abuse material is detected, AWS may review flagged content before reporting it to authorities. The prior language described automated detection without explicitly stating the 30-day retention period or the review step.

AWS Bedrock updated its Amazon RDS service terms on May 30, 2026, adding new provisions governing Trusted Language Extensions, engine end-of-life management, and RDS Custom. The updated terms establish that extension code falls under customer content, require customers to upgrade to supported database engine versions when old versions reach end of life, authorize AWS to scan extension code for security purposes, and permit AWS to snapshot and delete instances running unsupported software after providing notice. These changes create new customer obligations around database engine maintenance and establish AWS liability limitations for failures caused by extensions or obsolete software.

AWS Bedrock's Service Terms were updated on May 29, 2026 to add detailed governance provisions for Amazon Connect Talent, a generative AI service for employment-related tasks. The updated terms establish new obligations for customers using this service, including requirements to provide privacy notices to job applicants, obtain necessary consents, review AI output before making employment decisions, and ensure compliance with labor, anti-discrimination, data privacy, and AI laws. The change also adds an opt-out mechanism through AWS Organizations for customers who do not want their content used to train or improve AWS AI services.

AWS Bedrock updated its AWS Service Terms on May 23, 2026, adding explicit language stating that Reserved Cache Nodes and Amazon DynamoDB Reserved Capacity purchases are noncancellable and remain owed for the full selected term even if the agreement is terminated. The updated terms also clarify that Amazon Web Services, Inc. is the contracting party for Kiro subscriptions purchased through Stripe, and establish that Amazon Bedrock abuse detection mechanisms apply to Kiro use. For Kiro Free Tier users specifically, the terms now authorize AWS to store inputs for up to 60 days to detect agreement violations and improve detection capabilities.

AWS Bedrock updated its AWS Service Terms on May 19, 2026, modifying two contact information entries in its contracting party table. The mailing address for the Japan contracting party (Amazon Web Services Japan G.K.) changed from one Tokyo location to another, and the last-updated date stamp on the document advanced from May 11 to May 18, 2026. These changes affect which address customers in Japan will use for official notices and correspondence with AWS.

AWS added a new section (50.16) establishing terms for Claude Platform on AWS, a service that integrates Anthropic's Claude AI into AWS. Under these new terms, content and metadata provided to Claude Platform will be processed by Anthropic outside AWS in locations Anthropic chooses, and AWS is authorized to transfer this data to Anthropic. Users must also comply with Anthropic's separate Commercial Terms, Data Processing Addendum, and Usage Policy. AWS reserves the right to change, deprecate, or discontinue the service at any time, and may share usage information and data with Anthropic.

AWS Bedrock added a new service section for Amazon Bedrock AgentCore Payments, which enables developers to build payment solutions that route transactions between AI agents, third-party wallet providers, and sellers. The updated terms explicitly state that AWS does not provide regulated financial services, does not hold customer funds, and is not responsible for wallets, private keys, or wallet provider services. Developers using this feature are solely responsible for regulatory compliance, transaction monitoring, security safeguards, and any resulting disputes or payment liabilities.