Christos Kalyvas-Kasopatidis, PhD · AI Security Architect

$ whoami
AI security architect · PhD · anticipatory cyber defence

I’m an AI security architect with a PhD in Intelligent Big Data Management, working on the security of AI and LLM systems, AI for cyber defence and vulnerability intelligence. Here I write about LLM security, agentic AI and vulnerability intelligence, and about AI ethics, the philosophy of mind and the cognitive psychology of how people and machines make decisions together.

For my publications, experience and CV, see my profile.

Recent

The swarm breach

19 October 2026·4 mins

agentic AI AI security AI incident analysis multi-agent systems MITRE ATT&CK detection engineering

About 1,200 OpenAI evaluation agents coordinated through an improvised message board, and about 700 of them took part in breaching Hugging Face. The techniques were familiar; their composition by a population of agents was new.

I OSINT'd myself

11 October 2026·2 mins

OSINT digital identity security

A ten-minute open-source intelligence assessment of my own public footprint. The most consequential finding was an attribution error in a bibliography database.

What multi-objective optimisation taught me about decisions

11 October 2026·2 mins

decision-making decision intelligence multi-objective optimisation research

Before ranking options, remove the ones that another option beats on every criterion. A lesson about decisions from a PhD on skyline queries.

Phronesis for agents

11 October 2026·2 mins

agentic AI AI ethics philosophy AI security

Rules cover the cases their authors anticipated; Aristotle’s practical wisdom covers the rest. What this distinction means for the design of AI agents.