Yes. But how do you attract scientists if they can't publish their work at conferences, in journals, et c.?

By paying them lots of money and giving them interesting work. Obviously not everyone will make that trade but many will. There's plenty of work done in many areas of industry and government that aren't especially open and most of them probably don't pay nearly as well as Apple.

I recall that was an area of complaint about AI researchers a year or two ago. Valley companies were hiring them away from Academia and little research was being published as a result.

Principles are weak in the face of cash.

A friend bought his house with cash a couple of years ago with stock purchased through their EPP program as a retail employee from 2007-2011.

Fuck.. should have worked retail at Apple, self-studied CS and bought my own house instead of a $50k CS degree!

They had issues for a while in that area, actually. I hear that the ML people are now allowed to publish much of what they work on.

Yes, Siri has much more lenient policies for academic publications than other Apple departments, and other parts of Apple are also getting a little bit less strict. It was never entirely impossible to publish at Apple, but it used to be extremely difficult.

Once it became evident that this was a clear competitive disadvantage in ML hiring, policies evolved…

I don't think Apple is known for being particularly litigious when it comes to going after ex-employees (except in cases of fairly blatant industrial espionage, e.g. https://nationalinterest.org/blog/buzz/apple-concerned-china...).

In comparison, Apple's reaction to a prolific leaker was fairly mild, and left him far from unemployable (I happen to know what company hired him next), let alone homeless: https://www.macworld.com/article/1018725/workerbee.html