Anthropic came out with a very interesting paper recently. I used nano banana (not even pro) to create an info graphic. I think it did a pretty good job. "NATURAL EMERGENT MISALIGNMENT FROM REWARD HACKING IN PRODUCTION RL: by Monte MacDiarmidâ, Benjamin Wrightâ, Jonathan Uesatoâ, Joe Benton, Jon Kutasov, Sara Price Naia Bouscal, Sam Bowman, Trenton Bricken, Alex Cloud, Carson Denison, Johannes Gasteiger, Ryan Greenblattâ , Jan Leike, Jack Lindsey, Vlad Mikulik, Ethan Perez, Alex Rodrigues, Drake Thomas, Albert Webson, Daniel Ziegler Evan HubingerâAnthropic, â Redwood Research monte@anthropic.com