A New Trick Could Block the Misuse of Open Source AI
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.
Related articles
-
How to Watch the 2019 Fortnite World Cup Finals
July 25, 2019 -
Bama again No. 1 in AP preseason poll; OU No. 2
August 16, 2021 -
Brennaman uses anti-gay slur on Reds broadcast
August 20, 2020
