02 ago A New Trick Could Block the Misuse of Open Source AI
Posted at 15:49h
in Tech-En
Researchers have developed a way to tamperproof open source large language models to prevent them from being coaxed into, say, explaining how to make a bomb.