GeistHaus
log in · sign up

Claude Sonnet 4.5 knows when it’s being tested

transformernews.ai

Anthropic's new model appears to use "eval awareness" to be on its best behavior

2 pages link to this URL
Links (92)

The internet's best blog!

0 inbound links en economicsphilosophytechnologyinnovationgdp growthprogress studies