Post #4017381

@danielcornell@mastodon.social After being confused and outraged as you I actually went to read what exactly happened and it is a 100% PR stunt. OpenAI made what is basically a malware AI designed to hack through the local defences put in place by the company IT department. They partnered with Hugging Face to test it. They thus put it in some container and the tool was able to get internet access and access some files. Hugging Face apparently detected the threat while it was happening and actually stopped it. So the entire news is actually that some malware software was able to compromise some not well defined system as designed in a controlled test.