But I bet (probably well-founded) fears of drawing anti-trust or copyright infringement legal action has them hesitant to push that functionality on something a

But I bet (probably well-founded) fears of drawing anti-trust or copyright infringement legal action has them hesitant to push that functionality on something as prominent as the Search page. A relatively small company with no pre-existing monopoly position summarizing pages so you don't need to visit them is one thing. Google doing it (too much, and too well) is practically guaranteed to draw one or more credible challenges (again). This may leave a gap in which to challenge Google and win. But even the little company is likely to see some risky lawsuits as they monetize and grow. A lot's gonna hinge on whether telling a computer to do some math on content launders copyright as well as having a human learn something, then explain it, or synthesize something "new". And actually, it'll need to launder it much better, in the eyes of the courts, or else still be hopelessly vulnerable to legal attack—it's not like humans don't unintentionally infringe copyright, as the courts see it, pretty often, so if "AI" is held to the same standard it'll just be a mass-lawsuit-magnet. Uber for getting yourself sued.