AI is accelerating attacks and credential exposure. Security teams need a reliable path for outsiders to report what they find.
Corporate secrets are leaking at an alarming rate. GitGuardian’s 2026 State of Secrets Sprawl found 28.65 million new hardcoded secrets in public GitHub commits in 2025, up 34% year over year . AI is adding pressure from both directions: AI-assisted commits leaked secrets at roughly 2× the baseline rate , while leaked secrets associated with AI services grew 81% year over year .
As teams adopt AI tools, they create more tokens, API keys, and service identities, often faster than the governance around them can mature. More credentials in more places means more opportunities for one to surface somewhere it should not.
Security teams are rightly investing in preventing those exposures. But prevention will never catch everything. An outsider can discover an exposed credential, sensitive data, or another serious security issue before the affected organization does. When the first person to notice is acting in good faith, discovery only helps if the warning reaches someone with the authority to act.
Related: RemoteThreat Bets Security Teams Need to Test What Happens After Defenses Fail
That raises a less familiar security question: as your exposure grows, are you becoming any easier to warn?
Build the response path before the report arrives
A reporting channel only helps if there is a process behind it. For something like an exposed credential, that means knowing who validates the report, who owns the credential, and who can revoke or rotate it.
The workflow should be practiced before an incident. Validate findings using your own systems. Identify what the credential can reach. Revoke or rotate it, then audit the exposure window and look for downstream copies. If a production credential has not been changed in years, an active incident is a poor time to discover which systems depend on it.
This model is well established. The difficulty for many is making that path work in practice.
Make yourself easy to find
Researchers should not need personal connections to reach your security team. At minimum, publish a monitored security@ address, a security page, and a security.txt file at /.well-known/security.txt. The destination is as important as the file itself. Shared inboxes change owners, security pages go stale, and security.txt files expire. Someone should own the path and test it regularly.
The policy behind that path should also make room for unexpected findings. A company credential may surface in a third-party repository or service the company does not control. A useful disclosure policy lets an outsider report the finding without first deciding whether it fits a narrow asset list.
Related: Kiteworks & Citrix Incidents Show Challenges of Zero-Day Response
Do not make the bug bounty the only door
Bug bounty programs and vulnerability disclosure programs solve related but different problems. A bounty program defines what qualifies for a reward and often what researchers are invited to test. A disclosure channel gives outsiders a way to report something important whether or not it qualifies for a bounty.
A finding, like an exposed, active credential, can be out of bounty scope and still require immediate internal review. Security teams should know whether a third-party triager can close a report before anyone inside the company sees it, and whether a direct report gets handled internally or forces the reporter to start over on another platform.
Test whether a warning can get through
In May 2026, GitGuardian researchers found 844 MB of sensitive CISA-related data in a public GitHub repository, including still-valid credentials. Automated alerts had already gone out to the commit author, but the finding had not yet reached the people positioned to act. Once direct with CISA was established, the agency moved quickly, and the repository was taken offline within roughly 26 hours.
The lesson here is that detection and response are connected by a routing problem.
Run the test yourself. From outside your normal security workflow, can someone find a reporting route in under five minutes? How long until the report is acknowledged? How long until it reaches the right owner? Then try this same test with a harder case involving an asset you do not control or a finding outside normal bounty scope.
Related: SWIFT Banking & Government Middleware Enables RCE
AI is making both software and attacks move faster. Your warning path should be designed for the same reality.
Guillaume Valadon, Cybersecurity Researcher at GitGuardian
Want more Dark Reading stories in your Google results?
The State of Cloud Security: The Latest Challenges
The State of Cloud Security: The Latest Challenges
How Organizations Are Managing Incident Response
How Organizations Are Managing Incident Response
How Enterprises Are Developing Secure Applications
How Enterprises Are Developing Secure Applications
Inside RSAC 2026: security leaders reveal the risks redefining your defense strategy
Inside RSAC 2026: security leaders reveal the risks redefining your defense strategy
Essential News & Insights from Black Hat USA 2025
Essential News & Insights from Black Hat USA 2025
Building Cyber Resilience: Beyond Prevention to Recovery
Building Cyber Resilience: Beyond Prevention to Recovery
The Frontier AI Threat: Closing the Mobile Gap in Your Exposure Management Strategy
The Frontier AI Threat: Closing the Mobile Gap in Your Exposure Management Strategy
Static Analysis, Smarter Triage, Agentic Depth: A Practical AppSec Stack for AI-Driven Development
Static Analysis, Smarter Triage, Agentic Depth: A Practical AppSec Stack for AI-Driven Development
Effective Alert Triage: Reducing Noise and Finding Real Threats
Effective Alert Triage: Reducing Noise and Finding Real Threats
Cybersecurity Outlook 2027
Cybersecurity Outlook 2027
Europe's Multilingual Reality Exposes AI Security Gaps
Apple Reverses Age-Old Patch Policy to Keep Up With AI
Hand CVE Over to the Private Sector
Shutdown Sparks 85% Increase in US Government Cyberattacks
The full story
This article is one source in a clustered incident — the cluster page carries the summary, timeline and every other outlet covering it.
