What did 50 open source projects learn about security when AI started writing their code?

GitHub invested more than $500,000 in this round. The money moved through GitHub Sponsors as non-dilutive grants — no equity taken, no strings attached beyond measurable security outcomes. Seventy-one maintainers showed up. Ninety-two percent of projects finished the sprint with secret scanning, code scanning, protected branches, private vulnerability reporting, and Dependabot all enabled.

The program runs on a fixed rhythm. A three-week sprint kicks things off. Then maintainers stay engaged for twelve months with office hours, peer community, and two follow-up check-ins at six and twelve months. Each check-in unlocks another $2,000. The first $6,000 arrives during the sprint itself.

Training during the sprint covers three focus areas. Week one tackles foundations of open source security. Week two moves to threat modeling and secure coding. Week three addresses AI security and vulnerability management. The GitHub Security Lab designs and curates the curriculum. Security experts from GitHub and partner organizations deliver it.

OpenClaw joined Session 4 because it is GitHub's fastest-growing open source project. Its maintainers wanted a stronger security posture before the project scaled further. By the end, they had built an incident response plan, expanded GitHub security tooling, audited GitHub Actions workflows, and strengthened processes for identifying and responding to security issues. Their experience mirrors the broader cohort: specific risks differed, but every maintainer needed knowledge, tools, and expert support to secure software as AI changed how they built it.

Where the work happened across five layers

The fifty projects were grouped by the role they play in the software ecosystem.

Twelve sit at the intersection of AI, automation, data infrastructure, and machine learning. Caracal, Deep Agents, DocsGPT, LadybugDB, LangChain, n8n-MCP, Nasiko, ONNX, OpenClaw, PageIndex, Scenic, and Serena increasingly serve as foundational components for modern AI workflows and production deployments. As AI adoption accelerates, security improvements here establish stronger foundations for emerging AI ecosystems.

Seven projects handle build systems, supply chain, and release tooling. Browserslist, CycloneDX Python Library, Cucumber, golangci-lint, JReleaser, postcss, and Task help developers test, validate, package, release, and maintain software across diverse environments. Tools in this group influence everything from software bills of materials and release pipelines to code quality and testing automation.

Another seven define core programming languages, runtimes, and foundational libraries. Byte Buddy, core-js, FS2, Gleam, htmx, Pkl, Pyodide, and termcolor shape how software is written, configured, executed, and extended. Improvements at this layer flow downstream to thousands of applications and developer ecosystems. Security gains in foundational runtimes and libraries extend to the many tools and applications that depend on them.

Eleven projects shape the everyday experience of building, testing, collaborating on, and using software. Cheerio, Ciphey, CodeRunner, Hoppscotch, MapStruct, Python Pillow, Proyecto Respira, Readest, ToolJet, Vuetify, and Yjs serve as widely adopted utilities, applications, and platforms that appear throughout developer environments and application stacks. Together they support API development, low-code platforms, collaborative applications, content processing, and software delivery workflows.

A dozen projects form part of the internet's operational backbone. Actix-web, aiohttp, Apache Solr, Apache ZooKeeper, etcd, FastAPI, Haraka, Hummingbird, mimetype, Sniffnet, Starlette, and UAParser.js handle APIs, networking, search, messaging, service coordination, and distributed systems infrastructure relied on by organizations worldwide. These technologies sit on the critical path of modern cloud applications and internet services.

AI security did not replace the old work

AI-related security questions appeared across projects in every category — from machine learning infrastructure and agent frameworks to developer tools and internet infrastructure. At the same time, established security responsibilities did not go away. Maintainers still needed to manage vulnerabilities, secure dependencies, protect release workflows, and prepare for incidents. AI introduced new risks and increased the speed at which maintainers needed to understand and respond to them.

The lesson from Session 4 is stated plainly in the announcement: AI security is not evolving in isolation. It is becoming part of the broader practice of building secure software. As that shift continues, maintainers will need practical education, trusted communities, and expert support that can evolve with them.

What the numbers show across all sessions

Session 4 is not the whole story. Across all four sessions and follow-up periods through August 2026, 188 projects and 290 maintainers have participated across 42 countries. GitHub, Microsoft, and external funding partners have contributed $1.88 million distributed through GitHub Sponsors.

Participating projects have identified and disclosed 533 new CVEs. They have performed more than 1,500 Dependabot security updates. They have resolved more than 650 exposed secrets. During the last six months ending in July 2026, participating and alumni projects fixed 4,210 CodeQL alerts and blocked 119 secrets from being exposed.

Who paid for it and who helped run it

Funding partners for Session 4 included Alfred P. Sloan Foundation, American Express, Chainguard, Datadog, Herodevs, Kraken, Mayfield, Microsoft, Shopify, Stripe, Superbloom, Vercel, Zerodha, and 1Password. Ecosystem partners included Atlantic Council, Ecosyste.ms, CURIOSS, Digital Data Design Institute Lab for Innovation Science, Digital Infrastructure Insights Fund, Microsoft for Startups, Mozilla, OpenForum Europe, Open Source Collective, OpenUK, Open Technology Fund, OpenSSF, Open Source Initiative, OpenJS Foundation, University of California, OWASP, Santa Cruz OSPO, Sovereign Tech Agency, and SustainOSS.

Applications for Session 5 close August 24. Organizations can also apply to become Funding or Ecosystem Partners. The program structure remains the same: three-week sprint, twelve-month engagement, $10,000 per project tied to verified security improvements, and direct access to GitHub Security Lab expertise.