I'm *morbidly* fascinated by how the major tech firms are increasingly using "screenshots of what you're doing on your computer / browser" as part of new AI "agent" offerings
i.e. the agent takes a screenshot of your screen, and -- using image-recognition, text-recognition, LLM autcomplete, etc -- tries and automagically Do Stuff for you
On a technical level, it's a clever -- if hilariously kludgy -- hack
But wow, what a *gibberingly huge* security threat!