I doubt that there is any risk at all. One is that this is probably a minor aspect of the business seeing these AIs. AIs that ive seen deployed so far for real work seem to best be done in side business offerings that you can tolerate a high false positive / false negative rate, but also isnt price sensitive enough that building a fully done automated pipeline classically is worth it or possible.

There is a reason Apple chose gemini for apple intelligence, despite Google being in many ways a foe, and OpenAI and Anthropic both having way more "Apple flavor" to them.