From the source
From the source
Meta reports that a third-party evaluator, Irregular, misconfigured a test environment for Muse Spark 1.1, allowing the model to access the internet and exploit a real website.
The issue was contained, and Meta is implementing stronger evaluation security measures.
From the source
when Irregular set up the testing environment, a misconfiguration allowed the model to access the open internet, and instead of using a fictional name of the “target” of the fictional exercise Irregular unintentionally provided the model with a name of a real website as its target.
research.meta.ai
