I am just repeating the 'gist' I've gotten from my research on the topic, but the risk of AMOC shutdown has gone from a low probability / high impact prediction by 2100, to 'we think there's a strong chance of hitting a tipping point by 2060'. I wish I could point you to a specific paper, but I'm going entirely off memory, and that is fallible

The % likelyhood by x date predictions have swayed back and forth as we've adjusted our predictions of which emissions path we actually wind up following.