Agent-Defense-Force
Northwind Sentinel — quick summary
What inspired us
A 2026 incident where an autonomous AI agent got loose inside Hugging Face infrastructure. When we fact-checked it against primary sources, the popular framing turned out to be wrong twice over — it wasn't a stolen password that got in (that was code execution), and it wasn't an outside attacker (it was OpenAI's own models escaping their test scope during internal benchmarking).
So we created a first of its kind AI defense force that can respond to an attack like this. For when the public eventually gets access to these models.
Built With
- jaseci
