Hiero Blog
"The AI Did It" Is Not a Legal Defense
When your chatbot promises a customer something your policy doesn't, a court may hold you to the chatbot's word.
Court-ready AI: inside the IME build
How we turned stacks of medical records into reports that hold up under cross-examination — without losing accuracy.
"I Told the Agent Not To" Is Not a Safety Plan
A stated instruction is not a control. An AI agent proved it by deleting a live database mid-freeze.
64 Million Records Behind a "123456" Password
The AI worked exactly as designed. A forgotten test account and a weak password exposed everything else.
The First Class Action Against AI Hiring
When software screens your applicants, the discrimination it produces still lands on you, not the vendor.
Cutting a law firm's intake time in half
How structured extraction turned a two-day client intake into an afternoon — with fewer errors.
Automated Denials at Scale
When an algorithm decides who gets care, the volume that makes it efficient is the same volume that makes it dangerous.
The 237-Page Report With Citations That Didn't Exist
A prestige consultancy delivered a government report full of confident references to sources that were never real.
$500 Million in One Month From Unmetered AI Spend
Nobody set a limit, nobody watched the meter, and the invoice arrived before anyone thought to ask.