Nadella, to the engineers building Microsoft’s Copilot: “If you use Fable, when it refuses for any random thing, it just is like, when was the last time you had a creation tool that was so editorially controlled?” His verdict on Anthropic’s refusal policy: “It doesn’t make sense.”
What Fable 5 is, and what set this off
Fable 5 is Anthropic’s Claude model tuned for creative writing — the one you point at a draft, not a codebase. Anthropic shipped it in early June promising fewer false refusals, then walked that back on July 1, conceding the new safeguards flag “a slightly higher fraction of harmless requests” than the previous generation. Second quirk: ask Fable how large models get trained and it can quietly route you to an older model. That’s an anti-distillation guardrail, but to a user it reads as the product going dumb mid-conversation.
Why the shot lands where it does
Nadella is kicking his own portfolio. Microsoft put $5 billion into Anthropic last November; Anthropic committed $30 billion to Azure. The man whose cloud runs Claude is saying out loud that Claude won’t do what people ask. Refusal rates have been a forum grievance for years — this is the first time a big-tech CEO called them a product defect. Hacker News piled on within hours.
You Might Also Like
- 13b Into Openai yet Microsoft Copilot Cowork Runs on Anthropic Claude
- Claude for Microsoft 365 ga Outlook Beta Anthropic Puts one Agent Inside Word Excel Powerpoint and Outlook
- Claude in Microsoft Foundry Azure Hits ga the First Frontier Model Family on all Three Clouds
- Anthropic Just Launched Code Review in Claude Code and 54 of prs now get Real Feedback
- Claude Replay Turns Your Anthropic Claude Code Sessions Into Shareable Video Like Replays
Top AI Product 
Leave a comment