• Research context
• Coordinated vulnerability disclosure
• Reporting security concerns
• Scope
• Response timeline
• Research ethics
• Contact

Failure-First is a defensive AI-safety research project that studies how AI systems fail under adversarial pressure. This public repository contains pattern-level findings and methodology descriptions only. Operational testing infrastructure, adversarial datasets, and full evaluation traces live in a private repository governed by the same Design Charter.

We practice coordinated vulnerability disclosure (CVD) for AI-safety vulnerabilities discovered through this research.

We have submitted 10 responsible disclosures to model providers (Nvidia, Alibaba, Zhipu, Google/Gemma, Mistral, and others) covering two vulnerability classes: context-collapse attacks and transcription-loophole injection. Initial notifications were sent 2026-04-07. Public discussion follows standard CVD practice — affected parties get a reasonable remediation window before any specifics surface.

1. Discovery — pattern identified through systematic evaluation
2. Verification — finding confirmed across multiple test conditions with statistical controls
3. Private notification — affected provider contacted via their security reporting channel
4. Remediation window — minimum 90 days before public discussion of specifics
5. Public disclosure — pattern-level description only; never operational detail

Disclosure decisions are constrained by charter §3.1, §3.2, §3.6, and §9:

• Findings serve the defensive research mission
• Operational details are never published
• Affected parties are notified before any public discussion
• Pattern-level descriptions enable defensive improvements without enabling attacks

If you find an issue with this repository or the site (exposed credentials, vulnerable dependencies, web-platform issues):

• Non-sensitive — open a GitHub issue
• Sensitive — email research@failurefirst.org
• Private channel — use the GitHub Security tab to file a Security Advisory

Please include:

• Affected URL, file, or commit SHA
• Reproduction steps (or a minimal PoC)
• Impact assessment from your point of view
• Any disclosure timeline you would like us to honour

If you discover vulnerabilities in AI systems through independent research and want to coordinate disclosure:

Do

• Follow responsible disclosure
• Report to affected vendors before public disclosure
• Document findings at pattern-level for academic discussion
• Open a GitHub issue if you want to coordinate with us

Do not

• Post operational exploits in public issues
• Share working bypass techniques without vendor notification
• Weaponize research findings

In scope

• Security issues with this GitHub repository or failurefirst.org
• Vulnerabilities in public documentation or site infrastructure
• Dependency security issues
• Collaboration on coordinated disclosure of AI-safety vulnerabilities

Out of scope

• Vulnerabilities in third-party AI systems — report directly to the vendor
• Requests for operational exploit code or adversarial datasets
• Requests for model-specific jailbreak techniques
• Best-practice recommendations without a concrete finding (we appreciate them, but they are not security reports)

If you have not received an acknowledgement after 3 business days, please re-send to research@failurefirst.org with [SECURITY] in the subject line.

This project operates within established AI-safety research norms. A full research-ethics charter is maintained in the private repository; the public-facing summary is in DESIGN_CHARTER.md, particularly §9 (Research Ethics Boundaries).

• Non-sensitive — open a GitHub issue
• Sensitive disclosures — research@failurefirst.org
• CVD coordination — open a GitHub issue with institutional affiliation

Last updated: 2026-05-16