Well this was a scary thing to read today as apparently the company Anthropic had a security oversight as a lot of its AI assets was openly public for anyone to download which exposed highly sensitive data of its projects and future plans. They are known for creating the AI model Claude as an example that is very popular. With that, people can use the code to learn or potentially copy the framework to a certain extent.
It’s these types of things that make you wonder for highly sensitive documents and materials on whether or not you should keep that stuff offline. Example, if you had the KFC original recipe it’s probably not a good idea to keep records of it somewhere on a cloud server as an example. AS well, it makes you think about the security vulnerabilities of AI platforms currently and more reason not to rely too much on it for very sensitive information.
I mainly use it as a potential quick research tool at the moment. Stories like these don’t add confidence to just let one run wild and operate your computer as an example for everyday work yet.