Our goal is for AWS to be the most secure place to host any workload, and to support that we've been investing heavily in security across all of our services since AWS was founded more than two decades ago. Our AI services like Amazon Bedrock are built on this foundation and with the same focus. Bedrock offers customers world-class performance, security and privacy and a wide selection of models available anywhere. Last year we introduced the Bedrock Mantle with industry-leading privacy and protection for model weights. We often hear from customers that they want access to the latest models as soon as possible after their release and Bedrock delivers this along with the business features customers expect from AWS. We're excited that Anthropic's Claude Fable 5 models will once again be available to our customers on Bedrock starting tomorrow, and that they feature even stronger guardrails to prevent misuse.
When we release models, we don't just think about our responsibilities to our customers, but to the internet and society at large. The latest generation of frontier models, such as Anthropic's Claude Mythos have powerful new capabilities, especially in the area of cybersecurity. We were able to experience this first hand as part of Project Glasswing and are eager to get the Mythos-class models into the hands of defenders. As defenders, we have the opportunity to use these models to make the systems we all rely on materially more secure. But as we do, we must also ensure that we do not give our adversaries the visibility and capabilities that have been purposefully advanced, without giving companies, governments, and academic institutions the opportunity to protect their assets first. Achieving this balance is a key challenge for a wider model release, which is why we've been working closely with Anthropic and other industry partners on Project Glasswing to refine the roadmap for this new class of models. We all agree that preventing adversaries from gaining the ability to conduct in-depth vulnerability research is the most important goal of these security measures.
This is also an exciting time for AI, with new capabilities being introduced almost daily. We believe that making the capabilities of these advanced models available to all customers in a secure, privacy-preserving environment is critical to ensuring that they can reap the maximum benefits without creating security risks. It is important that new guardrails continue to be developed as we learn more about how well the current ones work and as new models are released. We will continue to iterate with our partners, deliver greater value, and respond to changes in the industry.
It is equally important to make sure that any problems with these models after their release are handled properly. Anthropic published a blog, Redeploying Fable 5, explaining how they think about the capabilities of this new class of models, and their commitment and SLAs for responding to issues reported to them. We appreciate Anthropic's transparency and collaboration in presenting this first architecture of the problem and the answers to the models that use the Internet, and we look forward to continued discussion throughout our field as we learn and refine it.
Our red AI team has worked with Anthropic to further improve Fable's defenses, and we believe its latest iterations result in a more capable model that further reduces the risk of adversary abuse. It brings the promise of more robust thinking power across multiple domains, without giving adversaries significant new security capabilities. When guardrails is activated, it automatically reverts to Opus 4.8, which is itself a world-class model that is already publicly accessible.
We appreciate Anthropic's partnership and commitment to defenders, and look forward to working with them and the rest of the industry to continue to make frontier models available safely and securely.
About the author
