AI turbocharges both development and attacks—software supply chains face a new defense crisis

The software development landscape is moving at an unprecedented pace. GitHub now processes roughly 275 million commits per week as of April 2026, compared to around one billion for the entire year of 2025. Meanwhile, GitHub Actions—the platform's automation tool—has scaled from 500 million compute minutes weekly in 2023 to 2.1 billion in a single week this year. But this explosion in output masks a deeper shift: the people writing code are no longer always human.

When AI agents make critical decisions#

Modern development workflows increasingly rely on AI coding assistants to generate entire modules, select dependencies, and structure applications. This means that non-technical teams in finance, HR, and business intelligence can now build software without traditional engineering expertise. The tradeoff is clear—developers no longer have direct visibility into which libraries, packages, and components their tools pull into a project.

This abstraction introduces a blind spot in software supply chain security. Humans provide high-level direction, but the AI agent decides how to accomplish the task, including which third-party components to integrate. According to security leaders, the people authorizing the work often never see the actual choices their AI tools make.

Attackers weaponizing the same technology#

Threat actors are leveraging frontier AI models with advanced cyber capabilities—such as Anthropic's Claude Mythos Preview and OpenAI's GPT-5.6-Cyber—to identify and chain together vulnerability chains that humans typically miss. A collection of medium- and low-severity issues that would normally sit low on a remediation queue can be transformed into a viable attack path when an AI agent threads them together strategically.

The speed advantage has become stark. Mean time-to-exploit has collapsed from 63 days in 2018–19 to an estimated minus seven days in 2025, according to threat intelligence from Mandiant. That negative figure means attackers can begin exploitation before security patches even exist.

Open source infrastructure under siege#

Modern applications depend heavily on open source components, making the supply chain itself an attractive attack surface. Campaigns like TeamPCP have successfully compromised widely used projects including Aqua Security's Trivy, injecting malicious code into trusted components that propagate downstream to thousands of dependent applications.