Open-source AI models vulnerable to criminal misuse, researchers warn

Hackers and other criminals can easily commandeer computers operating open-source large language models outside the ...