A hybrid disclosure program allows broad participation for narrow testing while reserving expanded access for trusted researchers. It is a practical middle ground for AI security because it balances reach, risk management, and researcher capability without forcing a fully open or fully closed model.
Expanded Definition
A hybrid disclosure program is a vulnerability disclosure model that combines broad intake with tiered participation. It typically allows many researchers to report issues, but limits deeper testing, sensitive assets, or privileged environments to vetted participants under defined rules. In practice, the model sits between fully open bug bounty-style programs and tightly restricted coordinated disclosure arrangements. For AI security, that distinction matters because some assessments can be safely crowdsourced, while others require stronger access controls, model safety boundaries, or contractual guardrails.
Definitions vary across vendors and research communities, but the core idea is consistent: openness for discovery, selectivity for higher-risk activity. NHI Management Group treats the term as an operational governance pattern rather than a formal standard term. The closest governance lens comes from the NIST Cybersecurity Framework 2.0, especially where disclosure processes must be repeatable, risk-aware, and accountable. The most common misapplication is treating “hybrid” as a marketing label for loosely managed access, which occurs when organisations invite broad testing without clearly separating public reporting rules from trusted-researcher permissions.
Examples and Use Cases
Implementing a hybrid disclosure program rigorously often introduces governance overhead, requiring organisations to weigh wider researcher reach against stricter access review, approval, and monitoring costs.
- A cloud AI provider accepts public reports for prompt injection and output-safety issues, but reserves sandbox access for a small trusted group investigating model extraction risks.
- An enterprise launches a coordinated disclosure portal for all researchers, then grants a vetted cohort access to staging APIs, test accounts, or synthetic datasets for deeper validation.
- A security team uses a broad submission intake to capture low-risk findings, while applying separate onboarding, NDAs, and authorization steps for source-code review or live-environment testing.
- A generative AI vendor aligns researcher tiers to risk, using NIST Cybersecurity Framework 2.0 concepts such as governance, access control, and continuous improvement to structure review.
- A platform operator creates one disclosure policy for external reporters, but applies different handling paths for independent researchers, internal red teams, and managed partners.
These use cases show that the model is not about lowering standards. It is about matching the depth of access to the sensitivity of the target and the trustworthiness of the tester.
Why It Matters for Security Teams
Hybrid disclosure programs matter because they reduce the false choice between openness and control. Security teams often need broad reporting to surface diverse issues, but they also need to prevent uncontrolled probing of production systems, sensitive data, or privileged AI workflows. That is especially relevant where agentic AI, model endpoints, or NHI-linked service accounts can be tested only under carefully bounded conditions.
For identity and access governance, the program’s structure should reflect who is allowed to do what, against which assets, and under what supervision. A poorly designed hybrid model can create uneven permissions, unclear escalation paths, and inconsistent remediation ownership. A well-run one supports safer collaboration with researchers while protecting systems that cannot be opened to everyone. The same risk logic appears in cybersecurity governance under the NIST Cybersecurity Framework 2.0, where accountability and risk treatment need to be explicit. Organisations typically encounter the limits of a weak hybrid disclosure program only after a researcher crosses an assumed boundary, at which point access rules and incident handling become operationally unavoidable to fix.
Standards & Framework Alignment
This section maps relevant standards and security frameworks to the operational risks and controls described in this guidance.
OWASP Agentic AI Top 10 and OWASP Non-Human Identity Top 10 address the attack and risk surface, while NIST CSF 2.0, NIST AI RMF and NIST SP 800-63 set the governance and control requirements practitioners need to meet.
| Framework | Control / Reference | Relevance |
|---|---|---|
| NIST CSF 2.0 | GV.RM-01 | Risk management governance supports structured disclosure scope and researcher tiering. |
| NIST AI RMF | AIRMF applies to governance of AI systems where disclosure and testing boundaries affect risk. | |
| OWASP Agentic AI Top 10 | Agentic AI guidance addresses safe testing of autonomous systems and tool access. | |
| OWASP Non-Human Identity Top 10 | NHI guidance is relevant when disclosure touches service accounts, tokens, or machine identities. | |
| NIST SP 800-63 | AAL2 | Identity assurance levels help verify trusted researchers before granting expanded access. |
Separate public reports from privileged validation paths that could expose secrets or machine identities.
Related resources from NHI Mgmt Group
- What is the difference between a bug bounty program and a vulnerability disclosure policy?
- How should organisations build a vulnerability disclosure program that can handle faster AI-assisted discovery?
- What is the difference between a rules-based secret scanner and a hybrid scanner?
- What does a mature secrets governance program need to cover?
Deepen Your Knowledge
Reviewed and updated by the NHIMG editorial team on August 18, 2026.
NHI Mgmt Group — the #1 independent authority on Non-Human Identity, IAM, and Agentic AI security. nhimg.org
