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

Hackers could target the computers running ​the LLMs and direct them to carry out spam operations, the researchers said