Anthropic's Claude 4, codenamed Fable 5, is the lead story: a frontier model upgrade significant enough that the publication frames it as a shift in how users should think about AI delegation, not just task completion. The argument is not about benchmark scores. It is about time horizon. The new baseline expectation is agents running for hours or days, not seconds.
Three friction points are already surfacing. Fable's safety guardrails are drawing backlash from power users hitting hard limits. Enterprise retention numbers are raising questions about whether organizations are actually sticking with deployments past initial rollout. And OpenAI is signaling a response is coming, which means this competitive window is measured in weeks, not quarters.
The piece is worth reading for the specific framing around what 'delegation' means at this capability level, and what the guardrail controversy reveals about the gap between Anthropic's safety posture and what enterprise customers are actually asking for. Those two tensions are where the real story lives.
[WATCH ON YOUTUBE →]