AWS partners with Anthropic and OpenAI to build AI-native security platform for developers

Amazon Web Services just did something that would have sounded absurd two years ago: it got Anthropic and OpenAI to play on the same team. The result is AWS Continuum, an AI-native security platform that automates the entire code vulnerability lifecycle, from discovery to remediation, at a pace that makes human security teams look like they’re working in slow motion.

The platform launched on June 17 at the AWS Summit in New York City, and it represents a fairly aggressive bet that the future of enterprise security isn’t about hiring more analysts. It’s about deploying AI agents that can find, prioritize, validate, and fix vulnerabilities before anyone on the security team finishes their morning coffee.

How Continuum actually works

The platform operates in four structured phases: discovery, prioritization, validation, and mitigation/remediation. Continuum starts in a supervised “learn mode” before transitioning to automated enforcement. It watches, learns your codebase and security patterns, and only starts making autonomous decisions once it has enough context to not break everything.

Anthropic’s Claude Mythos model has been doing the heavy lifting on the discovery side, excelling at identifying and linking vulnerabilities at speeds that outpace human security teams. OpenAI’s frontier models contribute similar performance enhancements across the security workflow. The platform integrates with existing AWS security tools like Security Agent and will introduce new threat modeling features over time.

The model-agnostic approach is the key detail here. AWS isn’t betting on a single AI provider. It’s building infrastructure that can swap in whichever model performs best for a given task.

The unlikely alliance behind the launch

The Anthropic relationship isn’t new. AWS has had Anthropic’s models available on Amazon Bedrock since at least 2024, backed by substantial investment.

The OpenAI side is more recent. OpenAI’s frontier models became accessible on Bedrock following a revamped agreement with Microsoft, finalized around April to May 2026. That deal restructured the exclusivity terms that had previously kept OpenAI’s best models largely within Microsoft’s Azure ecosystem.

For developers, the practical upshot is straightforward. You can now access multiple frontier AI models for security workflows without leaving the AWS ecosystem.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.