If "infosec people are not okay", that means that you have to account for the human element in your security analysis. You are not a consistent, trustworthy system, you are a sack of water and proteins.
Infosec processes aim at reducing this effect.
That's also AI companies attack surface.
AI companies have a profit motive to keep you exhausted, fearful, and demoralized, as it acts as a human DDoS against your health and emotional wellbeing. This encourages you to rely on them more and more to respond "at speed", fearful that the "exploits will get written anyways".
And yes, some of those are indeed and real vulns, they are a brief flash in the pan.
AI companies are using the exact same human DDoS fascism employs to reduce dissent and then encourage adoption via capturing the managerial class whom hasn't before been affected by neoliberalism.