Independent · open-source cybersecurity research lab & think tank
Psypher Labs
Quis custodiet ipsos custodes?
Who watches the watchers?
Old question, brand new problem. Right now, AI is becoming the thing that guards everything: it watches the networks, spots the attacks, patches the holes, and keeps the systems we all run on standing. Basically, the watchers keeping our world safe are turning into AI. Which sounds great, until you ask the obvious next thing: if AI is watching everything else, then who's watching the AI? What happens when the guard itself gets tricked, jailbroken, or turned? That's the whole game now. Someone has to watch the watchers, and that someone is us.
Mission
Psypher Labs is an independent, open-source cybersecurity research lab and think tank, a volunteer-run research collective. Our mission is to preserve and secure open source and bring people together to find solutions to some of the most challenging cybersecurity problems in the age of AI.
It's a space where everyone is welcome, a place for people to come together, share what they know, explore, and find a home, here in San Francisco and online.
Let's make the world a bit safer, one byte at a time.
Remain coherent for the word human to include us all and not exclude anyone.
What the lab builds
AI Security
Psypher AI Threat Assessor
Attacks the model, audits the infrastructure, and maps every finding to MITRE ATLAS, ATT&CK, and D3FEND.
Research
DPC-YOLO26
An NMS-free defense that suppresses adversarial patches while retaining real objects.
→ Code on GitHub: github.com/dreamsudo