See above.

What was this time frame?

See surveillance capitalism comment above.

The academic and corporate language is deliberately obfuscating. You and I have 'neural networks', machines don't. Artificial neural networks is almost as misleading.

Should the format match 'privacy'?

Building on my earlier comments about machines not having agency, I think the appropriate term is re-produce. I agree that with the best motivations we can reproduce care, but I don't ascribe ethical motivations to corporations or governments. I agree with Kate Crawford that power is the primary motivation for these large hierarchical institutions and their machines.

I added periods after citations for consistency, but am unclear whether there should be periods before too.

Does this category include AI alignment?

I think this is both the most accurate and least obfuscating term.

This is the AI alignment problem. https://spectrum.ieee.org/the-alignment-problem-openai
Connor Leahy, EleutherAI Cofounder, talks about alignment as THE challenge going forward.

Another way of obfuscating the outcomes of AI. Whether the model is open sourced, or not, the output has been statistically transformed through so many layers that auditing and transparency are impossible. Only the results can be questioned and that's where power is applied most effectively.

AI alignment

The Wiccan Rede is 'An' ye harm none, do what ye will'. Aleister Crowley's Law of Thelema: "Do what thou wilt shall be the whole of the Law"—is supplemented by a second, followup statement: "Love is the law, love under will." These two statements are generally believed to be better appreciated or better understood by considering a third statement: "Every man and every woman is a star."
https://en.wikipedia.org/wiki/Thelema

How can this be applied to the proprietary technologies that corporations deliberately mystify?