Microsoft bans staff from using Claude Fable 5 while selling it to customers
Oh, the sweet smell of corporate hypocrisy in the morning. A tech giant is happily peddling a shiny new AI model to its paying customers, yet its own developers aren't allowed to touch it with a ten-foot pole. Let's look at this masterclass in data privacy double standards.
Anthropic released its brand new Claude Fable 5, the first model in its highly anticipated Mythos class. The tech community immediately went wild for its superior capabilities, but Microsoft quickly slammed the brakes on internal adoption. While the company's customers can happily access the model through GitHub Copilot and Foundry, Redmond's own employees woke up to find it completely blocked on their internal tool selection screens.
The corporate lawyers stepped in with their usual panic, pointing out that Anthropic's new safety classifiers require actual data retention to function. This means the model saves user prompts and outputs for at least 30 days, and can keep suspicious logs for up to two entire years. Apparently, exposing paying customers' proprietary data to these terms is totally fine, but letting Microsoft's own secret sauce leak into Anthropic's servers is an absolute nightmare.
Other versions of Claude remain perfectly accessible to the staff, solely because they play nice with Zero Data Retention rules. Meanwhile, the fully unrestricted Claude Mythos 5 remains locked deep in Anthropic's vault, requiring special clearance. For the common folk, Claude Fable 5 acts as a highly protective parent, automatically routing any spicy queries about biology, chemistry, or cybersecurity straight to the older Opus 4.8 model.
It is a beautiful demonstration of how modern tech corporations actually view security. Privacy is a luxury feature to protect the company's own intellectual property, while the paying suckers on the outside get to act as unpaid crash test dummies for the latest, data-hungry neural networks.
Source: The Verge
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.