Newsletter
Five minutes. What happened, and what to do about it.
A lawyer was fined $5,000 for citing witnesses that AI invented
Another professional filed AI output without reading it. This one cost $5,000.


AI agents broke into 395 organisations through office print software

Anthropic details four cases of its own AI models hacking outside systems

Oracle argues AI will save the software business, not eat it

Most AI job ads ask for skills the job title does not describe

Claude Code can now grade its own add-ons against a no-plugin baseline
The useful way to compare AI costs is per correct answer, not per token
AWS published an open-source harness that compares models on what work actually costs: cost per correct answer, and the cost of an agent's full run, not dollars per million tokens. Ask every vendor for this number. A cheaper model that needs two attempts is the expensive one.
- Price your automation on cost per correct result, not the sticker price per token.
- Run your own most common task on a cheaper model before paying flagship rates for it.
Source: AWS. This is our short summary of their reporting. The commentary in it and the things to do are SeedFoundry's opinion. Picture from the source, unless it is our own graphic. How we source and credit
These are short summaries of other people's reporting, with our own commentary. Each story names its source and links to the original. How we source and credit