That seems like an easy statement to prove. How many bugs were there before AI vs after?
I may be wrong, but I would guess that you haven't seen any data to back up your statement and you're basing it on your perception based on social media posts.
You see a lot of clickbait articles where the author highlights a specific patch note or vulnerability and tries to tie that to AI. They're doing that to earn revenue because anti-AI posts get traffic... they're not trying to objectively inform you about the rate of bugs in Microsoft's products. Your perception is being skewed by selection bias.