Techsocial 101

Is your AI getting dumber?
Anthropic's latest safety update has triggered a massive 70% drop in coding performance for Claude Fable 5.
A new classifier is secretly rerouting difficult tasks to a weaker fallback model without telling users.
Developers are reporting that the AI they paid for is now failing at basic debugging.
Anthropic admits the trade-off is to stop a specific hacking technique but has no timeline for a fix.
This secret performance cut is leaving professionals with a significantly less capable tool.