GitGuardian SecDays San Francisco frames the credential layer for AI-era security

At a glance

What this is: GitGuardian’s SecDays San Francisco event centers on credential-layer security for AI-era software development, highlighting the scale of secret exposure and the governance gap it creates.

Why it matters: IAM, NHI, and security teams need to treat secrets exposure as a lifecycle and development-governance problem, not only a vaulting problem, because AI-assisted coding changes who creates credentials and how quickly they leak.

By the numbers:

• 28.7 million new secrets were exposed on public GitHub alone in 2025, the largest jump on record.

👉 Register for GitGuardian SecDays San Francisco on securing the credential layer for AI-era enterprises

Context

Credential-layer security is the discipline of controlling how secrets are created, exposed, used, and retired across software development and operational workflows. In this article, GitGuardian frames that problem around AI agents, developer tooling, and the growth of unsupervised application creation.

The governance gap is that existing secret-management models assume credentials are generated and handled by bounded human processes. When AI-assisted development and broad employee app creation enter the picture, the control point shifts from storage alone to the full lifecycle of machine credentials and secrets.

For IAM and NHI teams, the important question is no longer whether a secret was vaulted, but whether the organisation can govern the conditions under which it was created, copied, embedded, and reused.

Key questions

Q: What breaks when secrets are allowed into AI development workflows?

A: Secrets stop behaving like isolated credentials and start behaving like replicated content. AI tools can copy them into prompts, histories, logs, and generated code, which multiplies the cleanup surface and extends exposure beyond the original system. The practical failure is propagation, not just disclosure.

Q: Why do exposed developer credentials and secrets create such a high-risk path for attackers?

A: Exposed credentials matter because they bypass many defensive layers and give attackers immediate, legitimate-looking access. In cloud and development environments, a single token, key, or private credential can unlock repositories, workloads, or crypto assets, especially when standing access and weak monitoring are in place. Attackers prefer these paths because they are faster and quieter than exploiting software flaws.

Q: How do security teams know when secret sprawl is becoming unmanageable?

A: When they cannot confidently answer where each secret exists, which workloads depend on it, and how quickly it can be retired without breaking business services. If the answer requires manual archaeology across code, tickets, and pipelines, the sprawl is already beyond routine control.

Q: How should organisations govern AI agents alongside human identity and device access?

A: Organisations should treat AI agents as a separate identity class with their own entitlement boundaries, logging expectations, and approval model. Human IAM controls often assume interactive sign-in and review cycles, which do not fit autonomous or programmatic access. The safer approach is to define actor-specific policy and verify which access paths can be delegated without expanding trust unnecessarily.

Background and context

Why the credential layer becomes the control point in AI-era development

The credential layer is where authentication material meets software creation. In AI-assisted development, code generation can introduce secrets into repositories, build artifacts, and application logic faster than manual review cycles can catch them. That matters because secrets are not just storage objects. They are active credentials that confer access to cloud services, APIs, and internal systems. Once generated or copied into an uncontrolled workflow, they can persist beyond the developer who created them. The practical issue is not only leakage detection, but whether organisations understand which workflows are permitted to mint, transport, and reuse secrets.

Practical implication: Map where secrets originate in AI-enabled development and close the approval gap before credentials are embedded into code.

How AI agents change secret exposure patterns

An AI agent is a software entity that can make runtime decisions about actions and tools. In this context, the risk is that agentic systems may generate code, invoke services, or surface credentials without the same accountability chain used for human developers. That creates a different identity problem from classic developer mistakes: the credential may be produced, consumed, and propagated by a system that is not easy to recertify in a normal review cycle. If an agent can move from task to task with access to secrets, the organisation has to treat the agent itself as a governed non-human identity, not just as an application feature.

Practical implication: Classify AI agents that handle secrets as governed non-human identities and define who owns their access scope.

Why secret sprawl is now a governance issue, not just a hygiene issue

Secret sprawl happens when credentials proliferate across code, tickets, logs, chat, and build systems faster than they can be inventoried or revoked. The scale cited in the article shows that this is no longer a narrow engineering hygiene problem. It is an identity governance problem because each exposed secret expands the access surface and weakens assurance over who or what can authenticate. In practical terms, the organisation needs lifecycle visibility across issuance, distribution, exposure, and retirement. Without that, vaulting becomes a partial control over a much wider credential ecosystem.

Practical implication: Track secrets as governed identities across their full lifecycle, not as isolated values stored in one system.

NHI Mgmt Group analysis

Credential creation is now a governance event, not a side effect of coding: When AI systems and non-specialist employees can generate application logic, the act of creating a secret becomes part of identity governance. That expands the control boundary from storage and rotation into authorisation for credential creation itself. Practitioners should treat secret issuance as a governed workflow, not an incidental development artifact.

Secret sprawl is the operational symptom of broken credential lifecycle ownership: The article points to a world where secrets appear faster than they are discovered or removed. That breaks the assumption that credentials remain inside a manageable inventory long enough for standard review cadences to work. The implication is that lifecycle ownership has to be explicit across development, security, and platform teams.

The credential layer is the new identity perimeter for AI-era enterprises: If agents and employees can both create and propagate secrets, then access control is no longer centred only on humans or workloads. It is centred on where credentials are allowed to exist, how they are transported, and who can cause them to be issued. Practitioners should reframe identity governance around credential pathways rather than account lists.

AI-assisted development amplifies NHI risk because it normalises unsupervised secret propagation: The article’s scale signal matters because it shows that machine-created or machine-handled secrets are not rare exceptions. They are becoming a routine production pattern. That means NHI governance must move upstream into developer tooling, agent workflows, and code generation systems before exposure occurs.

Credential-layer governance: This is the right named concept for the article because it captures the shift from secrets as stored assets to secrets as governed identity material. Once credentials are created inside AI-enabled workflows, the decisive question becomes which systems are authorised to mint, move, and expose them. Practitioners should build policy around that pathway, not only around vault placement.

What this signals

Credential-layer governance is the right operating model for AI-era development because it shifts attention from storage to the full path a secret takes through tools, code, and runtime systems. That is where exposure now happens, and that is where policy needs to bite.

Secret sprawl becomes harder to manage once AI-assisted development normalises the creation and propagation of credentials outside traditional review loops. IAM and NHI programmes should therefore treat secret generation, copying, and reuse as governed events, not engineering leftovers.

For practitioners

• Inventory secret creation points Map every workflow that can mint or copy secrets, including AI-assisted coding tools, CI/CD steps, chatops, and developer sandboxes.
• Assign ownership for AI-handled credentials Define who owns secrets when an AI agent, coding assistant, or automation flow introduces them into code or runtime systems.
• Shorten exposure windows Reduce the time a secret can remain valid after it is introduced into a repository, build artifact, or prompt-driven workflow.
• Review developer-tool permissions Limit which tools can access, display, or copy secrets so that normal development activity cannot spread credentials into uncontrolled channels.

Key takeaways

• AI-assisted development is changing where secrets originate, which means credential governance now starts before a secret ever reaches a vault.
• The scale of exposure cited in the article shows that secret sprawl is a lifecycle problem with direct identity consequences, not a narrow cleanup task.
• Organisations that want better control need ownership, visibility, and revocation across the full credential path, including AI-driven workflows.

Key terms

• Credential Layer: The credential layer is the operational layer where teams track where secrets live, what they unlock, who owns them, and whether they are still valid. It sits across application, infrastructure, and identity controls, because a credential’s risk depends on its placement and its remaining access.
• Secrets Sprawl: The uncontrolled proliferation of sensitive credentials, API keys, tokens, passwords, certificates, across codebases, cloud environments, CI/CD pipelines, and configuration files. In 2024, over 50 million leaked secrets were found on the dark web.
• Credential Lifecycle: Credential lifecycle is the process of issuing, rotating, expiring, and revoking secrets, certificates, and tokens across their usable life. For non-human identities, lifecycle discipline is the core control that separates temporary access from persistent exposure.
• AI-assisted development: A development workflow where an AI system helps generate, refactor, test, or execute code inside the software delivery process. The security issue is not the assistance itself, but that the tool may operate in trusted contexts that also contain credentials, infrastructure access, and other sensitive assets.

What to expect at the briefing

GitGuardian's full event listing covers the operational detail this post intentionally leaves for the source:

• Session timing, agenda flow, and the in-person structure for San Francisco attendees
• The executive-facing fireside chat on governing the agentic enterprise and AI adoption
• Talks on the credential layer and where GitGuardian says AI agent and identity security now intersect
• Networking format and attendance details for security leaders who want the room-level discussion

👉 GitGuardian's full SecDays listing covers the agenda, speaker lineup, and attendance details for the San Francisco summit.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity security are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are building or maturing an identity security programme, it is worth exploring.