GitGuardian’s State of Secrets Sprawl report for 2025 reveals the alarming scale of secrets exposure in modern software environments. Driving this is the rapid growth of non-human identities (NHIs), which have been outnumbering human users for years. We need to get ahead of it and prepare security measures and governance for these machine identities as they continue to be deployed, creating an unprecedented level of security risk.
This report reveals an astounding 23.77 million new secrets were leaked on GitHub in 2024 alone. This is a 25% surge from the previous year. This dramatic increase highlights how the proliferation of non-human identities (NHIs), such as service accounts, microservices, and AI agents, are rapidly expanding the attack surface for threat actors.
NHI secrets, including API keys, service accounts, and Kubernetes workers, now outnumber human identities by at least 45-to-1 in DevOps environments. These machine-based credentials are essential for modern infrastructure but create significant security challenges when mismanaged.
Most concerning is the persistence of exposed credentials. GitGuardian’s analysis found that 70% of secrets first detected in public repositories back in 2022 remain active today, indicating a systemic failure in credential rotation and management practices.
Organizations may believe their code is secure in private repositories, but the data tells a different story. Private repositories are approximately 8 times more likely to contain secrets than public ones. This suggests that many teams rely on “security through obscurity” rather than implementing proper secrets management.
The report found significant differences in the types of secrets leaked in private versus public repositories:
This pattern suggests that developers are more cautious with public code but often cut corners in environments they believe are protected.
GitHub Copilot and other AI coding assistants might boost productivity, but they’re also increasing security risks. Repositories with Copilot enabled were found to have a 40% higher incidence rate of secret leaks compared to repositories without AI assistance.
This troubling statistic suggests that AI-powered development, while accelerating code production, may be encouraging developers to prioritize speed over security, embedding credentials in ways that traditional development practices might avoid.
In an unprecedented analysis of 15 million public Docker images from Docker Hub, GitGuardian discovered more than 100,000 valid secrets, including AWS keys, GCP keys, and GitHub tokens belonging to Fortune 500 companies.
The research found that 97% of these valid secrets were discovered exclusively in image layers, with most appearing in layers smaller than 15MB. ENV instructions alone accounted for 65% of all leaks, highlighting a significant blind spot in container security.
Secret leaks aren’t limited to code repositories. The report found that collaboration platforms like Slack, Jira, and Confluence have become significant vectors for credential exposure.
Alarmingly, secrets found in these platforms tend to be more critical than those in source code repositories, with 38% of incidents classified as highly critical or urgent compared to 31% in source code management systems. This happens partly because these platforms lack the security controls present in modern source code management tools.
Alarmingly, only 7% of secrets found in collaboration tools are also found in the code base, making this area of secrets sprawl a unique challenge that most secret scanning tools can not mitigate. It is also exasperated by the fact that the users of these systems cross all department boundaries, meaning everyone is potentially leaking credentials into these platforms.
Further exacerbating the risk, GitGuardian found that leaked credentials frequently have excessive permissions:
These broad permissions significantly amplify the potential impact of leaked credentials, enabling attackers to move laterally and escalate privileges more easily.
While organizations increasingly adopt secret management solutions, the report emphasizes these tools alone aren’t enough. GitGuardian found that even repositories using secrets managers had a 5.1% incidence rate of leaked secrets in 2024.
The problem requires a comprehensive approach that addresses the entire secrets lifecycle, combining automated detection with swift remediation processes and integrating security throughout the development workflow.
As our report concludes, “The 2025 State of Secrets Sprawl Report offers a stark warning: as non-human identities multiply, so do their associated secrets—and security risks. Reactive and fragmented approaches to secrets management simply aren’t enough in a world of automated deployments, AI-generated code, and rapid application delivery.”
