Following the launch of its Bedrock service for creating generative AI applications, Amazon has introduced a new enterprise offering for CodeWhisperer, its AI-powered tool for generating and recommending code.
This fresh offering, aptly named CodeWhisperer Enterprise Tier, brings forth enhanced capabilities that enable companies to seamlessly integrate their internal code repositories and resources with CodeWhisperer. This integration empowers CodeWhisperer to provide tailored code recommendations to developer teams.
Amazon explains in a press release that administrators utilizing the CodeWhisperer Enterprise Tier can establish a connection between CodeWhisperer and a private code repository. Subsequently, CodeWhisperer will familiarize itself with the repository and adapt its suggestions accordingly. Customizations can be managed via a console, where administrators can also access evaluation metrics, estimate the performance of each customization, and selectively deploy them to developers to prevent wide exposure of proprietary code.
Amazon notes that customization will initially be available in preview as part of the CodeWhisperer Enterprise Tier.
CodeWhisperer Enterprise Tier represents Amazon’s latest expansion of CodeWhisperer, which was initially launched in late June as a component of the AWS IDE Toolkit and AWS Toolkit IDE extensions in response to GitHub Copilot.
In April, Amazon made CodeWhisperer available to developers for free without any usage restrictions. During the same month, CodeWhisperer Professional Tier was introduced, which included features like single sign-on with AWS Identity and Access Management integration and higher limits on security vulnerability scanning.
It’s worth noting that code generation services have not been immune to the controversies surrounding other generative AI technologies. In many instances, the models supporting code generation services like CodeWhisperer are trained on copyrighted code or code under restrictive licenses. Vendors assert that fair use, a doctrine in U.S. law allowing the use of copyrighted material without prior permission from the rights holder, protects them, whether the models were developed knowingly or unknowingly using copyrighted code. However, not everyone agrees with this viewpoint.
Aside from vendors’ liability concerns, some legal experts argue that code generation services could potentially expose companies to risks if they unwittingly incorporate copyrighted suggestions from these tools into their production software.
To address these concerns, Amazon has implemented a feature in CodeWhisperer that highlights and optionally filters the license associated with functions resembling snippets from its training data. The company is also committed to not utilizing customer-specific customizations for future CodeWhisperer model training and assures customers that it won’t store or log customer content when handling requests from a developer’s IDE.
