Why Microsoft and Other Anthropic Customers Held Off On Using Claude Fable

Anthropic’s latest AI is so good at coding tasks that many firms are https://www.theinformation.com/articles/anthropics-mythos-security-powerhouse-budget-buster?rc=geicgp">willing to pay top dollar to use it even as Anthropic effectively raises prices.

But some customers are holding off over cybersecurity concerns.

In releasing its latest model, Claude Fable, Anthropic started a new policy of holding onto any data that customers feed into the model for 30 days to ensure malicious actors aren’t using the model for illegal or harmful ends.

That means companies that use Fable to write or edit employees’ code for developing or overhauling applications are essentially signing up to store their data with Anthropic for a month. 

In some cases, Anthropic stores the data in separate cloud servers from the ones customers have configured to use Fable. Amazon Web Services, which powers servers for Anthropic and also sells the startup’s models to cloud customers, told customers on Tuesday that “once you opt into data retention, your data will leave AWS’s data and security boundary.”