AI EXPERIMENTS / SECURITY PROVING GROUNDS
Curiosity.
With a dark side.
An AI security lab for exploring what’s possible—and understanding what can go wrong. This October, follow the frog into the unknown.
Question assumptions. Test boundaries. Keep learning.
From prompts to packets, every boundary is worth exploring.
01 / EXPLORE THE GROUNDS
Ten doors. Plenty to discover.
Pick a discipline. Find the risks, the questions, and a practical place to start.
01AI & automation
Prompt injection, tool permissions, retrieval boundaries, and trustworthy agent workflows.
Explore the topic
02Cloud & infrastructure
Asset inventory, configuration drift, workload identities, and cloud attack paths.
Explore the topic
03Governance & compliance
AI inventories, risk decisions, control evidence, and accountable lifecycle management.
Explore the topic
04Identity & access
Conditional Access, privileged access, authentication, and identity lifecycle hygiene.
Explore the topic
05Data security
Data classification, retrieval authorization, secret handling, and AI data exposure.
Explore the topic
06Threat security
Threat modeling, adversary behaviors, attack paths, and safe validation.
Explore the topic
07Endpoint management
Device inventory, secure configuration, patching, and endpoint protection health.
Explore the topic
08Application security
Secure defaults, dependency hygiene, API authorization, and software supply chains.
Explore the topic
09Network security
Segmentation, controlled egress, name resolution, and useful network telemetry.
Explore the topic
10Security operations
Detection engineering, hunting, investigation, and incident response practice.
Explore the topic02 / FIELD NOTES FROM THE LAB
Less theory. More flashlight.
Three hands-on guides to help you investigate, evaluate, and document what you find.
The haunted context window
Can an untrusted document change what your assistant is allowed to do?
Ghosts in your tenant
Use report-only evaluation to understand a policy before it affects access.
Signals in the dark
Turn a process-hunting question into a repeatable investigation.
Practice guides for your own test environment. Estimated times are starting points.
03 / THE REFERENCE SHELF
A few trusted spellbooks.
Go deeper with the people maintaining the frameworks and the platforms. These references open on their official sites.
04 / ABOUT GRIBBITE
A proving ground for curious minds.
GRIBBITE brings AI experimentation and security practice into the same room. The focus: understand the system, test the boundaries, and leave with something useful.
Explore identity, endpoints, cloud, data, and the new questions that AI introduces across all of them. The Halloween costumes are seasonal. The curiosity stays.
Independent lab notes. Practical questions. No magic guarantees.