# AWS partners with OpenAI and Anthropic for AI security collaboration The era of 'autonomous security' in code protection begins - **Source:** [VentureBeat](https://venturebeat.com) - **Published:** 2026-08-10 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/3effec0d-678c-464e-aeea-aa0d0afedfed ## AWS's major security move AWS announced integration of its Continuum code security platform with OpenAI's Codex and Anthropic's Claude Code. This strategic move embeds AWS security directly into competitors' development environments. Simultaneously, AWS expanded its Security Hub Extended marketplace with supply chain protection, adding Chainguard and Socket as partners. These steps are part of AWS's strategy to lead enterprise software security in the AI era. ## Massive vulnerabilities uncovered by Mythos Anthropic's Claude Mythos Preview, announced in April, demonstrated unexpected cybersecurity capabilities during testing. The model discovered thousands of zero-day vulnerabilities across all major operating systems and web browsers. Over 99% of these vulnerabilities remain unpatched. The median time from vulnerability discovery to weaponized exploit has dropped from 771 days in 2018 to under 4 hours in 2024, projected to fall below 1 hour by end of 2026. ## Vulnerability backlog raises alarms AWS VP Chet Kapoor revealed that Mythos has increased the vulnerability backlog fivefold. Organizations' vulnerability discovery rates now far exceed their remediation capacity. With the global cloud infrastructure market exceeding $143 billion per quarter, AWS developed Continuum to address this critical challenge. ## Vision for autonomous security AWS is shifting from traditional 'telemetry, storage, query, human dashboards' to 'telemetry, context, reasoning, and agent actions.' This vision is encapsulated in the phrase 'autonomous security at machine speed.' Continuum represents the implementation of this forward-thinking approach to cybersecurity. ## Continuum's four-phase system Continuum operates on an 'agent-team loop architecture.' The discovery phase employs multiple AI models to scan code. Prioritization evaluates findings against actual business risk. Validation creates reproducible exploits in isolated sandboxes to confirm exploitability. Remediation offers tested solutions including network config changes, policy adjustments, or code patches - all with human oversight options. ## Why prioritization matters While organizations typically have capacity to address 100 vulnerabilities, systems like Mythos can identify 2,000. Continuum's prioritization system determines which 100 vulnerabilities demand immediate attention. Kapoor describes this as 'one of our biggest value adds,' helping customers focus limited resources effectively. ## The validation process Continuum creates executable exploit code in isolated sandboxes to verify each vulnerability's exploitability and potential impact (blast radius). This process covers both customers' proprietary code and third-party open source dependencies, providing comprehensive security validation across the entire codebase. ## How remediation works Continuum delivers tested solutions including network configuration changes, policy updates, or code patches - all validated in sandbox environments before recommendation. The system offers flexible autonomy levels, allowing organizations to choose their preferred balance between automation and human oversight. ## The commercial model Customers pay AWS a single price for Continuum. AWS absorbs the underlying AI model costs, optimizing which model to use at each scanning phase. 'Customers purchase Continuum, period,' says Kapoor. 'We optimize model selection because frankly, GPT Cyber excels at some tasks while Mythos shines at others.' ## Collaboration over competition AWS's partnership with rivals OpenAI and Anthropic represents a strategic shift. Kapoor rejects labeling them as 'competitors,' insisting they're 'partners.' AWS believes working with multiple model providers is essential, as 'everybody is going to leapfrog each other over time' in the rapidly evolving AI landscape. ## Industry implications AWS's move signals a redefinition of security in the cloud and AI sectors. Traditional security approaches prove inadequate against AI-powered threats. Systems like Continuum aim to provide proactive protection at the point of code creation. This trend will likely gain momentum in coming years as cybersecurity paradigms evolve. ## The future security architecture Continuum may become one of the first successful implementations of AI-powered autonomous security. Its success could dramatically reduce the time between vulnerability discovery and remediation, playing a pivotal role in preventing cyberattacks. In the AI era, effective security requires not just defense, but also prediction and automation. ## Sources covering this story - https://venturebeat.com/security/aws-continuum-integrates-with-openai-codex-and-anthropic-claude-code-in-major-ai-security-push — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._