US Secret Weapons Supplier Requirements Baffle Microsoft Boss

Microsoft CEO Satya Nadella has sharply criticised the content restrictions on Anthropic’s Fable artificial intelligence model, describing the limitations as nonsensical. He made the remarks to engineers working on the company’s AI Copilot. ‘If you use Fable, when the software refuses something random, it’s like when was the last time you had a creative tool that was so editorially controlled?’ Nadella said. ‘It doesn’t make sense,’ he added. Anthropic explained on a support page that users would receive responses from an older version when asking about certain topics. The rejections have also drawn criticism from a number of social media users. Fable 5 was announced in early June, with Anthropic stating the AI model was designed to reduce false positives for blocked requests. Three days after launch, Anthropic cut off access to Fable, citing compliance with US government export control directives. Access was restored on 1 July 2026, with the company saying it had added new safeguards to the model. ‘The new safeguards will flag a small fraction of harmless requests compared to Fable’s previous safeguards,’ Anthropic said. The restrictions placed on Anthropic’s AI products stem from a deteriorating relationship with the US government. Anthropic was previously the only AI trusted for daily operations by the US government. However, relations soured after Anthropic refused to allow the military to use its models for surveillance and fully autonomous weapons systems. In retaliation, the Trump administration placed Anthropic on a national security blacklist and ordered the suspension of exports of its Mythos and Fable models to various countries worldwide, citing national security concerns.