Akrites: The Latest Attempt to Protect Open-Source From AI Attacks Has Arrived - DevOps.com

Sign up for our newsletter!<br>Stay informed on the latest DevOps news

Akrites, a new Linux Foundation initiative backed by many of the world’s largest tech and financial firms, is the industry’s latest attempt to get ahead of AI‑accelerated software supply chain risks by hardening critical open source projects before attackers can exploit them.<br>On June 25, the Linux Foundation unveiled Akrites, a coordinated industry program designed to find, fix, and responsibly disclose vulnerabilities in open-source software exploited by AI-based attackers.<br>It’s not the first such effort. But Akrites may be the most comprehensive. One such initiative is Chainguard’s Athena coalition, which seeks to repair open-source flaws before attackers can exploit them. Another is IBM and Red Hat’s Project Lightwell, which has similar goals.<br>These two, however, seek to provide safe code and a platform for managing compliance, SBOMs, and governance across heterogeneous open‑source supply chains. Akrites’ mission, on the other hand, is to give the industry “one coordinated way to fix vulnerabilities upstream before they’re exploited.” In this approach, maintainers stay in control. Its focus is squarely on open-source software projects and ecosystems rather than on any one enterprise’s stack.<br>The reason all of these projects are appearing now is simple. Frontier AI models have proven they can scan large codebases and find exploitable bugs in minutes. How bad is it? At the United Nations Open Source Week conference, Linux Foundation CEO Jim Zemlin said, "The mean time to exploit a vulnerability in software is now negative seven days." In other words, by the time you find out there’s a security hole in your program, the hackers have had a week to exploit it.<br>It used to be that discovering and weaponizing serious flaws required deep expertise. Now, anyone with access to a top-level AI LLM can find security holes.<br>So it is that the Linux Foundation is positioning Akrites as a way to move defenders together, rather than leaving maintainers and individual vendors to fight a rising tide of machine‑generated vulnerability reports alone.<br>The effort launches with an unusually broad coalition of cloud providers, AI labs, financial institutions, telecoms, and security vendors. Founding commitments come from Amazon Web Services, Anthropic, Chainguard, Cisco, Citi, Endor Labs, Ericsson, Google, IBM, JPMorganChase, Microsoft, GitHub, NVIDIA, OpenAI, RapidFort, Red Hat, the Rust Foundation, Sonatype, Vodafone, and Zscaler. You may have noticed that the companies I mentioned previously as having their own initiative are backing Akrites as well.<br>That said, as Mike Dolan, the Linux Foundation’s SVP of Legal, noted, "the space is crowded, and the track record is mixed. What I think is genuinely different here is narrow: one coordinated process, so open source maintainers face a single partner instead of a hundred separate reporters. It’s not to say everyone will go through Akrites – many may not, but those who do help reduce the noise for maintainers."<br>All that AI-generated noise has been a real problem. On the plus side, AI reports today are far more accurate. On the negative side, that’s still an enormous number of bugs.<br>Moreover, vulnerability disclosure for open source is often anything but coordinated. Maintainers are flooded with overlapping or conflicting reports from multiple companies, governments, and independent researchers. Different teams may privately discover the same issue, develop competing patches or forks, and send them upstream via separate channels. Even when everyone is acting in good faith, the result can be noise, fatigue, and fragmentation.<br>Akrites is intended to replace that ad hoc model with a single, predictable entry point:<br>A shared Security Incident Response Team (SIRT) will serve as the trusted coordination partner for open-source maintainers, receiving vulnerability reports and managing remediation across projects.<br>A standardized, confidentiality‑first Coordinated Vulnerability Disclosure (CVD) process will rely on the familiar ecosystem of CVE identifiers, traffic light protocol (TLP) markings, common weakness enumeration (CWE), CVSS severity scoring, EPSS exploit prediction, SSVC decision‑support, and VEX statements to track exposure and communicate risk.<br>Fixes will flow back to each project’s original home on the maintainers’ terms, with Akrites explicitly committing not to “fork and fragment” open source as part of its response.<br>The idea is to give maintainers one clear signal—validated vulnerabilities and well‑tested, coordinated fixes—rather than a flood of duplicative requests.<br>As Dan Lorenc, Chainguard’s CEO and co-founder, remarked, "The software supply chain is only as strong as the upstream it draws from, and we see how thin that layer really is. As AI finds more vulnerabilities, the industry will rush to patch them....