Cybersecurity

This is a direct quote from the article:

There has been a lot of research pointing out the security risks related to the use of Pickle file serialization (dubbed “Pickling” in the Hugging Face community). In fact, even Hugging Face’s documentation describes the risks of arbitrary code execution in Pickle files in detail.

In other words, there's a known vulnerability, it's documented, it's ignored and now it's been exploited twice.

Wow .. shocked .. is not a word I'd use to describe this situation.

Fuck around and see what happens .. seems more apt.

Without reading the article, as a practicing AI Engineer here’s a couple of easy best practices:

• Use only the .safetensors files, the format is engineered specifically to allow only specific information necessary for NN frameworks
• Don’t use the ‘trust_remote_code=True’ parameter when serving your models without due consideration of the source of the model.

We've known pickle files have been unsafe for like three years at this point and people are still using them?