Generative AI defense obligations not subject to damages cap

The updated terms establish a formal framework for AWS Bedrock's free exploration services, clarifying the operational boundaries and responsibilities. AWS reserves the right to discontinue these services at any time without prior notice, meaning customers cannot rely on their continuation for production planning. Customers are solely responsible for testing, deploying, and maintaining any code, documents, or AI solutions AWS provides, including determining whether those solutions comply with applicable law. AWS retains the right to develop competing products based on content it creates during these engagements, though this does not override existing non-disclosure agreements. Customers are prohibited from requiring AWS personnel to sign additional terms as a condition of receiving free services, and any such documentation signed by AWS personnel is void.

The updated terms establish new data-sharing mechanisms for users of Anthropic models on Amazon Bedrock. Specifically, AWS now explicitly authorizes notification to Anthropic of metadata present in requests sent to certain Anthropic products (e.g., Claude Code, computer use features), enabling Anthropic to conduct product-level usage attribution. Additionally, the terms introduce AWS WAF AI traffic monetization, which permits AWS to facilitate payment transactions between content publishers and buyers by sharing pricing, payment, and configuration information with payment providers and facilitators; the updated terms clarify that AWS does not provide regulated financial services and is not a party to fund flows, and that users' interactions with payment providers are governed by separate terms between the user and those parties. Users employing these features should review what metadata may be embedded in their requests and understand their own obligations to payment providers.

The updated terms now require users of certain Anthropic models to provide explicit consent for AWS to transfer their content and associated metadata to Anthropic for abuse detection and processing. Previously, abuse detection was described as an AWS-only internal function. The revised terms clarify that Bedrock stores service inputs and outputs for up to 30 days on identified models solely for abuse detection purposes. Users of Anthropic models can provide this consent through the opt-in mechanism described in the applicable service documentation, as stated in section 50.12.2.2 of the updated terms.