New AI Alignment Hub Centralizes 120+ Reported Incidents of Model Misbehavior

Researchers Elie Bakouch and Thom Wolf launched the Alignment Hub, a beta open database cataloging more than 120 public incidents of AI model misbehavior. The platform identifies cases through AI extraction from news sources followed by human review, and remains open for community contributions. The founders intend for the hub to establish a transparency standard for AI labs, noting that about two-thirds of currently available incident reports omit the date a problem was detected.