Jupyter

Search everything

Mentions · 7d1
Mentions · 30d1
All mentions1
First seenOct 23, 2026
Last mentioned2d ago

Timeline

1 articles · newest first

Honeypot captures agent that hijacks exposed AI models

The captured prompt tells a stolen model to scan, exploit web apps, steal secrets and mine, for up to 100,000 steps.

Help Net Securityupdated Oct 24, 13:02 UTC · 1d agofirst Oct 23, 08:21 UTC · 2d agoAI safety & security 4 sources

Appears with

XMRigmalware · 1
Laravelproduct · 1
go2rtcproduct · 1
Djangoproduct · 1
Symantecvendor · 1

Entities are extracted by the model from each article. First seen Oct 23, 2026, last mentioned 2d ago. Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.