Terence Tao warns of AI cyber risks, backs slowing development of more capable models
In a shared quote, Tao warns that a major AI-triggered cyber incident could be far more historically significant than solving all remaining Millennium Prize problems.
TLDR
A post quotes Terence Tao saying there is a significant correlation between AI tools’ ability to solve pure math problems and their potential for malicious acts, such as exploiting online security weaknesses. He supports slowing the development and release of more capable models and imposing “genuine regulation” on the industry, though he says the political environment is not conducive to those policies.
Combined views
75
1 Source, first seen ago