SecurityBrief US - Technology news for CISOs & cybersecurity decision-makers
United States
GitGuardian launches programme to warn on leaked credentials

GitGuardian launches programme to warn on leaked credentials

Fri, 20th Nov 2026 (Today)
Mara Sugue
MARA SUGUE News Editor

GitGuardian has launched an Enterprise Good Samaritan Program to warn companies about exposed credentials linked to their systems. The scheme is open to organisations whether or not they are GitGuardian customers.

The move comes amid patchy enterprise responses to credential leak warnings, even as attackers become faster at finding and abusing exposed secrets.

Since spring, GitGuardian has linked 3,913 high-impact leaked credentials to 1,532 companies and contacted each organisation, usually reaching at least five people. Only 35% responded, it said.

That gap matters because notices sent to individual developers do not always lead to exposed credentials being revoked. About 90% of valid secrets remain active a week after the developer responsible for the leak has been alerted, according to GitGuardian.

The new programme is intended to route warnings to security teams and other internal contacts who can act on them. GitGuardian monitors public sources such as GitHub and Docker Hub for exposed credentials, then uses automated analysis and human review to determine whether a leak could provide access to company systems or data.

Growing volume

GitGuardian has detected more than 83 million exposed secrets on public GitHub since 2018 and sent 11.9 million alerts to 5.9 million developers. Its 2026 State of Secrets Sprawl report found that 28.65 million new secrets were exposed on public GitHub in 2025, up 34% year on year.

Its systems now screen more than 1 million public incidents each month and investigate more than 100,000 in depth. Human experts review the most serious findings before affected organisations are contacted with details on the validity of the credentials, their permissions, and the systems or data that may be at risk.

GitGuardian pointed to an incident involving the US Cybersecurity and Infrastructure Security Agency, or CISA, to illustrate why escalation beyond automated alerts can be necessary. In May, its researchers identified a public GitHub repository containing CISA-related data and valid cloud credentials.

Nine automated alerts sent to the person who created the repository did not resolve the exposure. After researchers escalated through CERT/CC and established contacts, the repository was taken offline within 26 hours of the human escalation, GitGuardian said.

AI pressure

GitGuardian framed the programme as a response not only to the scale of the problem but also to the speed at which attackers can now move. Earlier this month, Microsoft documented threat actors using agentic artificial intelligence tools to search public GitHub repositories for exposed credentials.

In one case cited by Microsoft, an AI agent moved from identifying a leaked secret to extracting data from Azure Key Vault in five minutes. That example adds urgency to efforts to shorten the time between exposure, warning, and revocation.

"For years, we've warned developers, but they're not always the ones who can revoke the secret. Our expertise combined with AI now helps us identify at scale the organizations behind serious exposures, and our researchers work to reach the people who can respond," said Éric Fourrier, Co-Founder and Chief Executive Officer of GitGuardian.

"Attackers now have the ability, with AI offensive harnesses like ARTEX, to use these credentials to target companies efficiently. Those companies deserve to know, whether they're our customers or not. And if a company has no clear way to receive a warning, the warning doesn't help. Every company needs a door for these reports, and someone behind it," Fourrier said.

Call for channels

Alongside the launch, GitGuardian is urging companies to create clear, monitored routes for receiving external security disclosures. Organisations should maintain channels such as a security.txt file and a staffed security inbox so warnings reach teams able to investigate and act.

The programme operates under disclosure principles that include timely notification, non-destructive verification, transparency, traceable research activity, and persistent follow-up, according to GitGuardian. Organisations that receive notifications do not need to sign up to the service or become customers to obtain details of an exposure.

The issue reflects a broader challenge for large organisations, which often have decentralised development activity, multiple code repositories, and inconsistent ownership of credentials. In such settings, a warning sent to a single developer or an unattended mailbox can easily be missed while the exposed secret remains valid and usable.

GitGuardian said the expanded programme aims to reduce that lag by getting disclosures to the people responsible for revoking credentials and containing any related access. Without a monitored point of contact, even accurate and timely warnings may fail to prevent abuse, it said.