Is Claude getting dumber? Daily first-hand reports of quality drops, silent fallbacks to weaker models, usage cuts and classifier false positives — plus the trackers, benchmarks and official responses that confirm or contradict them.