Skip to main content ↑ ↓
Topic: A New Trick Could Block the Misuse of Open Source AI (Read 39 times) previous topic - next topic

A New Trick Could Block the Misuse of Open Source AI

A New Trick Could Block the Misuse of Open Source AI

[html]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.[/html]

Source: A New Trick Could Block the Misuse of Open Source AI (http://ht**://www.wired.c**/story/center-for-ai-safety-open-source-llm-safeguards/)