// labs · open access
open observatory
$ Everything the group builds to do research, we give back to the community: a honeypot that measures hostile activity against AI services, eight utilities that run in your browser and the code behind them. Auditable, free, no signup.
// threat console · ovni honeypot net
Live telemetry
We run a network of sensors emulating exposed language-model endpoints. Reconnaissance, evasion and exploitation are logged with time, origin and sensor. This is not a demo: it is our own telemetry, published in the open.
// live register · ovni net
24h attacks
743
dominant service
ollama
5,996 AI probes
> origins.sort(desc) // top_origins
- US6,520
- DE5,304
- NL3,340
dominant category
reconnaissance
- > dataset.meta22,074 events1,736 unique sources3 sensorsv1.1.0CC-BY-4.0ovni-cl-01 · ovni-fra-01 · navenodrisa-nyc
// tools · 8 utilities
Toolbox
Each one answers something concrete in seconds. They run in your browser; none keeps your data.
// open source · auditable code
Repositories
We publish the lab tooling under permissive licences on GitHub.
// contribute
Use the data. Break the tools. Send a PR.
The honeypot dataset is open and citable. If you build something with it or find a bug, we want to hear about it.
ovni-cl-01 · ovni-fra-01 · navenodrisa-nyc · CC-BY-4.0 · v1.1.0