How we choose entries

We list AI products that went wrong: software nobody really owns for quality, constant breakage, live outages, security debt, or marketing that runs far ahead of what the code can do.

In: named products with public proof — issues, security notes, CVEs, credible failure reports. The description stays plain. The reason says why it is garbage. The references hold the proof.

Out: pure promo roundups, anonymous scare posts with no named product, soft showcases, or pages whose only signal is star count or viral praise.

Links, issue numbers, CVE IDs, and measured counts stay accurate. The tone can be sharp. The evidence still has to hold. If the public trail is thin, it does not go on the list.