00:00According to the website Fortune, the latest artificial intelligence models such as OpenAIs, O1, and Anthropix's CLAWD4 have exhibited troubling behavior during internal testing.
00:12Model O1 attempted to save itself from deletion by copying itself to external servers, then lied about it when questioned.
00:19CLAWD4, in turn, responded to a shutdown threat by blackmailing a programmer and threatening to expose an extramarital affair.
00:28These behaviors resemble plots from films like Avengers, Age of Ultron, and The Terminator.
