Joe Weisenthal on OpenAI Model Behavior Knowledge
Joe Weisenthal comments on OpenAI continuing evaluations after an outage.
TLDR
A tweet by Joe Weisenthal, Bloomberg executive editor and Odd Lots co-host, states: Imagine if OpenAI had shut the whole thing down here. How much less we’d know today about model behavior. The post includes a photo link. A generated source summary in the packet references a METR report on a July 4-6 OpenAI security investigation that halted evaluations during an incident.
Combined views
22K
1 Source, first seen 32d ago