
Microsoft does not want its employees to use Mythos-class Fable 5, here is why
Microsoft has reportedly restricted the use of Anthropic's first widely available Mythos-class AI model, Claude Fable 5. But the reason behind this may not be what you think.

Microsoft is one of the biggest tech companies out there with thousands of employees all across the globe. The tech giant has been pushing for AI tools in the workforce in hopes of increasing productivity. So, on the surface, you may think that Microsoft would be one of the many companies willing to use Anthropic’s Claude Fable 5 – the first Mythos-class model that is available to the public. However, reports indicate that this is not the case.
As per a report from The Verge, Fable 5 is not available for Microsoft employees in the company’s internal version of GitHub Copilot. This comes despite the fact that Microsoft has rolled out the Mythos-class model to customer versions of GitHub Copilot and Microsoft Foundry.
Keep in mind that Microsoft is said to have recently cancelled Claude Code plans for its employees with reports indicating that high costs likely played a part in that decision.
Why is Microsoft limiting use of Claude Fable 5?
But for Fable 5, costs are likely not the reason for restricted access. Though the Mythos-class model does cost twice as much as the Opus family of models. The report states that Microsoft is limiting employee access to the Fable 5 because of Anthropic’s data retention rules for the Mythos-class model.
Under Anthropic’s new policy, prompts and outputs generated through Mythos-class models are retained for 30 days to support the company’s safety classifiers, which are intended to monitor novel attacks and jailbreak attempts.
Anthropic says that the data is deleted after 30 days and is not used for model training. However, prompts or outputs flagged for safety investigations, legal purposes or possible usage policy violations may be retained for up to two years.
That shift has raised legal and compliance questions for enterprise users. Previous enterprise AI offerings typically provided zero-retention options, allowing conversations to disappear once they ended.
As per the report, Microsoft’s legal teams are reviewing Anthropic’s policy, with concerns centred on customer data, and confidential material. It is not yet clear whether the model will be cleared for broader internal use.
However, other Claude models are still available to Microsoft employees internally because they operate under zero data retention rules.
On the internet, some users joked that this news could be detrimental to Microsoft employees who were looking forward to using Fable 5 to vibe code – using AI to write code. One person shared a video of a person going back home and wrote, “Microsoft devs on their way home after realizing they can’t vibe code with Claude Fable 5 at work.”
Another person posted a clip from the film Prisoners where the actor Jake Gyllenhaal is breaking things in his office. The person added, “Microsoft employees have realised that they'll now have to really work (some of them don't know how).”
On the other hand, one user claimed that this may just have been a bad excuse as Microsoft may not want to pay millions in token use. The user commented, “Bullshit excuse. They just don’t want to pay $50 per million tokens.”



