Not AI.
The max generation time for the current models is around 5-8 seconds before coherency is lost. The details of the vestments, the movement of the cassocks, the consistency of the crowd and background are too good to be artificial. More likely, the there's an issue with the encoding and frame rate of the video that gives it a slightly choppy/slow-motion look to it. It might also have been recorded with a variable frame rate due to the low light in the chapel.