Microsoft CEO Satya Nadella criticized restrictions on Anthropic’s Claude Fable 5 model during an internal meeting with engineers working on Microsoft’s Copilot products, CNBC reported July 16.
Mr. Nadella described the model as “so editorially controlled” and said its restrictions “don’t make sense,” according to the story. Microsoft declined to comment to CNBC and Becker’s, and Anthropic did not immediately respond to the news outlet’s request for comment.
Five things health system leaders should know:
1. Some requests are automatically sent to an older model. Anthropic said Fable 5 checks every request and may reroute flagged prompts to Claude Opus 4.8. The categories include offensive cybersecurity, most biology, chemistry and life sciences requests, attempts to distill the model’s capabilities and a narrow set of frontier AI development tasks.
2. The restrictions could affect legitimate healthcare work. Anthropic acknowledged that the broad safeguards may intercept benign requests involving medical imaging and diagnostics, clinical and diagnostic healthcare questions, biotech documentation and basic biology education. Users receive a notice when a request is switched and can turn off automatic switching; API customers must opt in and configure the feature.
3. The criticism comes despite deep ties between the companies. Microsoft committed to invest up to $5 billion in Anthropic in November, while Anthropic committed to purchase $30 billion in Microsoft Azure computing capacity. Microsoft has also integrated the technology behind Claude Cowork into Microsoft 365 Copilot. Fable 5 is available as an optional preview model in Copilot Cowork, though it is off by default and administrators can disable Anthropic’s model family.
4. Fable’s availability has already shifted in response to government action. Anthropic suspended the model June 12 after the U.S. government imposed export controls, then restored global access July 1 after the controls were lifted. The company redeployed it with updated safeguards that it acknowledged could flag more benign coding and debugging requests.
5. For health systems, the dispute underscores the importance of documenting which model produces an output, testing fallback behavior and revalidating clinical, research and cybersecurity workflows when vendors change their underlying models or guardrails.
Editor’s note: Becker’s reached out to Anthropic for comment and will update the story if the company responds.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
