Open Secure AI Alliance launches to defend open-source software from AI-accelerated attacks

The Linux Foundation launched the Akrites initiative on June 25, 2026, alongside a founding cohort that includes Anthropic, OpenAI, AWS, Microsoft, Google, IBM, and NVIDIA. What the alliance actually does Akrites is built around two concrete mechanisms. The first is a shared Security Incident Response Team, or SIRT. The second is a single Coordinated Vulnerability Disclosure process built on established standards including CVE and CVSS: when someone finds a flaw in widely-used open-source code, there is one agreed-upon playbook for reporting it, triaging it, and fixing it. The initiative is backed by the Alpha-Omega fund, which provides initial financial support and is structured to accept additional contributors willing to put in either engineering resources or capital. At launch, the alliance published one pointed benchmark: fewer than 5% of recently surfaced open-source security issues had been patched as of June 25. Why AI changes the math on software security Where a skilled researcher might have spent weeks auditing a codebase to surface a critical vulnerability, modern AI models can do a version of that work in minutes. Akrites’ founding participants published an open letter...