Image this ...
Big Banks, Airlines, and Nuclear stations use Citrix or some other remote login client to let their employees log into their Virtual Cloud clients over VPN so they can control what their employees do and prevent things like data leaks.
However, the employee's BYOD has been compromised by bad actors who have gained access to Recall and now can not only see what goes on since breaking in but can Recall all the past screens/data of that computer, including the work they do for the bank, airline, or nuclear station. They see not only customer data but also gain knowledge of how those systems work and find more vulnerabilities way down the architecture chain. They see what other users may also have access to and then target their BYOD as well. And so on it goes, going right throughout an organisation without anyone knowing (as they have not used any of that information ... yet).
Now, when they hit these places, the companies react, but due to the internal people who fire-fight, these situations are also very much compromised as well (that's the clever part of the bad actors), which actually opens up more of the organisation to the bad actors. Essentially, Recall has given them way too many vectors to attack an org, and it just keeps giving more and more.
This whole approach is a massive security risk. I suspect they think the AI will prevent all this, but it won't as the bad actors will target that first as the best way to stay in once they are in.
At the end of the day, we need humans to make the decisions and store the key information in their heads. The simple reason is that while both Humans and Computers can be compromised, it's harder to keep Humans comprised in that way than Computers. As Humans make mistakes and can get found out, therefore their state of being completely compromised changes from that to a lesser one; they find it hard to break other Humans as well, and eventually and usually soon, they get discovered. Once computers get broken, then they are broken; they are then used to break into more, and you may not discover it until it's too late!