This week we present the first of two episodes on dynamic memory allocation.  We were going to talk about garbage collection, but realized we should probably discuss allocators first.  Turns...

Frog and Toad face the most modern of challenges

This week we present the first of two episodes on dynamic memory allocation.  We were going to talk about garbage collection, but realized we should probably discuss allocators first.  Turns...

Empowering everyone to build reliable and efficient software.

All modern language models sometimes act like they have emotions. What’s behind these behaviors? Our interpretability team investigates.

A middle ground between the cybersecurity and AI safety communities

Read the declaration and add your name.

Anthropic embeds imperceptible watermarks in Claude text and C2PA metadata in files, worldwide. What it means for API output, code, and detection.

Practical math for the working programmer.

This story is wild. The short version: OpenAI were running a cybersecurity test against an unreleased model, with the model’s guardrail features turned off. Rather than solve the test, the …

YouTube video by WIRED Investigations

This episode started with the question: who first had the idea to connect computers together to make a network?  We explore JCR Licklider's ideas about an  "Intergalactic Network", some of t...

Why & how we rewrote Bun from Zig to Rust

In this episode we discuss a couple of recent papers on just-in-time development and agentic systems and how we think they suggest trends that will affect software development for years to come. ...

This week we were joined by our friend Doug Burke, who runs a company and has been working on his "digital life workout partner", a voice-activated LLM tool. Doug's story is interesting because he doe...

In this episode we invite back our friend Bob Farzin to discuss our personal experiences with using teams of agents to write software, and we try to parse out the experiences we're seeing on the inter...

11 posts tagged ‘ai-security-research’. Using AI tools to help find security vulnerabilities.

Some recent articles about research on hash tables made us realize we probably didn't know enough about hash tables, one of the fundamental data structures in the biz.  We talk about the history ...

Today we are launching the beta of EmDash, a full-stack serverless JavaScript CMS built on Astro 6.0. It combines the features of a traditional CMS with modern security, running plugins in sandboxed W...