I spent the last few days documenting real AI failures — built a free searchable database anyone can contribute to

AI systems fail constantly — hallucinations, bias, harmful outputs, fabricated citations. But there's no single place where these failures are documented, analyzed, and searchable by researchers, journalists, and policymakers.
So I built one. It's called AIFAIL.LOG.
Every time someone submits a failure, AI analyzes it instantly — identifies the root cause, the pattern it represents, and the risk level. All data is public and searchable.
Here are 5 real documented failures already in the database: [list the 5 seed entries from your tracker]
Link: [ https://aifaillog.netlify.app/ ]
If you've witnessed an AI fail badly — submit it. The more real cases we document, the more useful this becomes for safety research.