Microsoft CEO Satya Nadella called out Anthropic in front of his own engineers last Wednesday, telling staff that the restrictions built into Anthropic's high-end Fable AI model do not make sense. The remarks were made at a staff meeting with engineers working on Microsoft's Copilot AI software, and a copy of Nadella's comments was provided to CNBC.
"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?" Nadella told the engineers. "It doesn't make sense."
Microsoft declined to comment on the remarks after the meeting. An Anthropic spokesperson did not immediately respond to a request for comment.
The criticism targets a specific pattern in how Fable handles certain user requests. When end users ask Fable about some aspects of creating large-scale models, among other topics, Anthropic might send responses from an older version of the model, according to an Anthropic support page. Some users have publicly called out the rejections on social media.
Anthropic's history with Fable this summer has been turbulent. When the company announced Fable 5 in early June, it said it was working to reduce false positives for blocked requests. Three days after the release, Anthropic cut off Fable access entirely to comply with a U.S. government export control directive. The company restored access on July 1, saying at that point that new safeguards would flag a slightly higher fraction of harmless requests than the previous version did.
The public criticism is notable because Anthropic is a significant Microsoft partner and client. In November, Microsoft announced a $5 billion investment in Anthropic. As part of that deal, Anthropic agreed to spend $30 billion on Microsoft's Azure cloud. This year Microsoft also unveiled Copilot Cowork, a business productivity assistant that draws on Anthropic's models. Anthropic's Claude Code software development tool has separately become widely used among programmers.
Despite that partnership, Nadella has been steering Microsoft toward a broader model strategy. He has argued that companies should be able to cost-efficiently develop custom models and draw on internal data without letting it flow out to other entities, including companies in the business of building models. In a Sunday blog post, Nadella invoked Palantir CEO Alex Karp, who said on CNBC that technical organizations want to know they own the means of production.
"It can't be that there are only two companies in the world with token capital, and everybody else is renting it," Nadella told the engineers at the meeting.
Microsoft's Foundry service currently offers developers access to more than 11,000 models, including some from Anthropic and OpenAI. Nadella's broader push reflects a shift among executives toward cost-efficient models that can handle software development and other tasks inside companies, rather than relying exclusively on the most well-funded labs.
Investors have followed Microsoft's AI strategy closely. Shares have fallen 17% so far this year, while the Nasdaq Composite has gained 11% over the same period. On Thursday, one day after Nadella's staff meeting, Chinese startup Moonshot AI announced an open-source model it said surpasses recent releases from both Anthropic and OpenAI, adding more pressure to the competitive landscape.
