Anthropic's New AI on AWS Is Smarter but Keeps Your Data Longer

Anthropic, the company behind the Claude chatbot, just launched Claude Fable 5.1 on Amazon Web Services (AWS, Amazon's cloud platform). This is the newest version of their most advanced AI model. It is designed for hard, multi-hour jobs like writing software, solving engineering problems, and doing deep research. The key upgrade is that it shows better judgment. That means fewer confidently wrong answers, and when it doesn't know something, it says so instead of making things up.

For an everyday user, the practical benefit is that an AI like this can act like a very patient assistant. It can take a large project, plan it out, use different software tools, fix mistakes on its own, and then give you a finished document or working code. Anthropic says it's especially useful in finance, accounting, healthcare, and scientific research.

However, there's a serious trade-off. Because Fable 5.1 is so capable, Anthropic has labeled it a "Covered Model." In plain English, that means Amazon will keep a copy of your prompts and outputs for up to 30 days, and a human reviewer might look at them. This is to catch abuse, but it's a real privacy issue for companies with sensitive data.

For big businesses, there's an option called Enterprise Frontier Safeguards. This lets eligible customers use the model with zero data retention, meaning Amazon and Anthropic won't hold onto your information. That option is currently available until the end of 2026. Later this year, even stronger protections will let companies run the AI entirely inside their own cloud infrastructure, with their own encryption keys, and only automated monitoring, not human review.