Apple tries to market its product as privacy-focused, yet the privacy of their new AI features is so bad they don't meet EU standards? Is that the message here?

It's the inverse problem. EU wants anyone to be able to install a different AI agent onto their phone with the same access as Siri. Apple says "no- we need time to figure out how that would work, we want other agents to meet the same privacy standards of PCC/on-device that Siri uses". Which EU said no.

I don't think there's a clear good guy/bad guy here.

I am not very sympathetic to apple here, but I think the legislation in question is more to do with competition than data privacy.

This conflict applies to many tools that require high privileges:

* If you allow the user to grant those privileges to third-party applications, they can grant it to applications that abuse it, resulting in security and privacy risks. You might even be blamed for allowing them access (e.g. the famous Cambridge Analytica scandal).

* If you don't allow the user to do that, third-party tools won't be able to serve those needs, which can be considered anti-competitive preferential treatment of your own tools.

The interoperability of their AI doesn't meet EU standards.

You've got to be delusional if you think the EU has a good grasp of AI.