Observed arrival · 2026-11-04
Prompt Mortem: an autopsy journal for AI failures
An independent journal records generative-AI experiments as case files, pairing prompts with outputs, model names, and reported compute costs.
- For
- Readers tracking generative-AI failures and costs
- Worth noticing
- The homepage reports eight disasters and USD 51.72 wasted to date; one featured case lists a USD 41.65 cost.
Field notes
Entries use numbered case files, with model names, reported costs, prompt text, and outcome summaries visible in the supplied homepage extract. The featured video experiment compares Seedance variants using shared prompts and references; its description also notes additional takes that changed the setup. The homepage labels its running total as money wasted, making the project's accounting part of the editorial frame rather than a performance claim.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue