When a new potentially-hazardous asteroid is discovered, it's normal for the probability of impact to go up a couple times before abruptly plummeting to zero, as the radius of uncertainty shrinks until it shrinks past the Earth. See https://en.wikipedia.org/wiki/Torino_scale#/media/File:Apoph....
A 3% chance of impact right after discovery with the initial error ellipse isn't all that unusual; it will almost certainly be revised to 0 with more observation (and if it's not, you'll hear about it).
A 3% chance of impact right after discovery with the initial error ellipse isn't all that unusual; it will almost certainly be revised to 0 with more observation (and if it's not, you'll hear about it).