The idea is good but even with ai measures minors can easily fake an ai face id verification. I don't think it's possible to prevent 100% of cases.
But hopefully if this actual results in the ai tech improving at preventing a over whelming majority of kids faking their ages it would be extremely rare. If not.. the amount of fines tech companies might risk facing could potentially bankrupt them if too many kept sneeking past the ai and the gov acted against Google for that. the fines for that type of content isn't usually a small matter.