The fact that there is no "user override"[0] feature in any of these security processors is blatantly obvious evidence that they are designed to control and restrict first and foremost. I have read mjg's other posts on the topic and have no reason to believe he is arguing in bad faith but I'm still not convinced one bit.
I wouldn't call it bad faith, but more like furthering the industry narrative.
Unfortunately a lot of intelligent individuals are perfectly content to help the corporations and governments tighten the nooses on everyone, including themselves, in return for $$$. They've convinced themselves that they are doing good.
[0]: https://www.eff.org/wp/trusted-computing-promise-and-risk