Microsoft CEO Satya Nadella has reportedly criticised the restrictions built into Anthropic’s latest AI model, Fable, saying the chatbot’s tendency to refuse certain user requests is difficult to justify. As per a CNBC report, Nadella made the remarks during an internal meeting with engineers working on Microsoft’s Copilot AI platform. He questioned whether an AI assistant should be so tightly controlled in the first place.
As per the report, Nadella pointed to the instances where Fable declined to answer prompts that users considered harmless. He argued that it felt unusual for the creative AI tool to be editorially controlled to such an extent.
Anthropic introduced Fable 5 in June with updated safety measures for reducing harmful outputs while minimising false positives. However, the company later acknowledged that its revised safeguards can still block some legitimate requests. The startup has also faced criticism from users on social media over refusals involving AI development and technical topics.
Also read: OpenAI enters hardware market with Codex Micro: What is it and how much it costs
Nadella also stated that Microsoft wants to make AI models more accessible instead of concentrating computing resources in the hands of a few companies. He reportedly said it made little economic sense for only a small number of AI firms to control access to large-scale AI infrastructure.
So far, the company offers over 11,000 AI models via its Azure AI Foundry platform, including models from Anthropic, OpenAI and other developers, alongside its own in-house models. The company has been actively developing its AI technologies despite investing billions of dollars in partnership with Anthropic and OpenAI.Nadella also reportedly admitted that Microsoft’s consumer and enterprise Copilot products should have been unified earlier, following the company’s recent organisational changes.