Writing
Thoughts on security, AI, and the bits of the industry I think are worth explaining properly.
10th May, 2026 ยท Edited 24th May, 2026
How I Accidentally Ended Up Helping Secure OpenClaw
How a security engineer accidentally ended up working full-time on OpenClaw, and what securing a fast-moving AI project actually looks like.
Read article24th July, 2024 PST 2024
Synthetic Trajectory Generation Through Convolutional Neural Networks
Jesse Merhi, Erik Buchholz, Salil S. Kanhere
A reversible trajectory-to-CNN transformation that lets convolutional GANs generate synthetic location data, evaluated against an RNN baseline across two real-world datasets.
Read on arXiv10th May, 2024
LLMs are Secure... for Now
How AI and LLM security risks look today, where they are overstated, and where agentic systems make the future harder.
Read article