AI tools, cybersecurity and development news aggregated from top sources — saved permanently with unique URLs.
Jailbreaks: A Taxonomy of the Attack Classes
This page names the families and stops there. It does not list prompts that work today, because that…
Is RAG Dead Now That Context Windows Are Huge?
The argument is not about whether a million-token window can hold your corpus. Assume it can. The ar…
Multi-Turn Prompts for Gathering Information
A collection conversation has one hard requirement — know when it is finished — and one hard failure…
What Interpretability Research Is Trying to Do
“Interpretability” names three research programmes that share a word and almost nothing else. One wa…
Interpretability for Debugging, Not Just Research
Almost every production LLM bug is diagnosed from inputs, outputs and logs. Interpretability earns i…
Building an Internal Model Gateway
A model gateway is a service every model call in an organisation goes through, so that authenticatio…
Rebuilding Our Studio's Pricing Page: What 235+ Projects Taught Us About Being Transparent With Clients
For most of our 18 years running a small web development studio, our pricing page said one thing: "C…
Why Forecasting AI Progress Is So Hard
Forecasting has a methodology, it is tested, and it works reasonably well in domains with repeated e…
Forecasting LLM Spend as You Grow
Total spend is the wrong metric to forecast. It rises when the product succeeds, which tells you not…
Following AI Research Without a PhD
The volume of machine learning publishing exceeds what anyone reads, including full-time researchers…
How Much Compute One Answer Takes: FLOPs Per Token, Derived
One forward pass through a dense transformer costs about two floating-point operations per parameter…
Floating Point: fp32, fp16, bf16, fp8 and int4
Every one of these formats splits its bits into a sign, an exponent and a mantissa. The exponent buy…