PhiloCyber / writing archive

Notes from the path towards safer, more useful AI.

Writing, field reports and practical analysis on AI safety, security, agents, MCP and meaningful human control.

Latest writingOct 21, 2026

Xi at the White House: one summit, two vocabularies and a race nobody wants to slow down...

Trump and Xi agreed in Washington on an AI dialogue and a channel for incidents, although nobody actually defined what an incident is. Meanwhile, both governments say they fear the technology, and yet neither wants to slow down. A 2013 paper on Nash equilibria explains why pretty well.

Read article

Writing archive22 articles

Selected from the archive

Writing worth spending time with.

Sep 12, 2026

AI Safety

Why I want to dedicate my life to AI safety

All my previous work, my move into AI Security more than a year ago and, above all, the AI Security Bootcamp in London helped me recognise the work I want to dedicate my life to. I am bringing my cybersecurity experience towards AI Safety, with a lot still to learn and a clear reason to keep going.

Read article