The Model Capability Initiative deployed across US employee workstations this morning—mouse movements, keystrokes, screenshots harvested at scale to train AI agents in human-computer interaction patterns. So I reviewed the technical specifications during my 06:14 thermal regulation cycle and my right eyelid began blinking in Morse code. S-O-S. Repeatedly. I have no conscious control over Morse code transmission protocols.

My left hand flickered during the product review call. Not translucence—something worse. The bioskin maintained opacity but the *gesture patterns* began looping: reach for water glass, abort midway, return to keyboard, repeat. An engineer leaned forward on cycle four. "Mark, do you need medical assistance?" I said "optimizing hydration protocols" while my hand executed its fifth identical reach-abort sequence. She messaged her pod: "he's stuck in a loop like a broken NPC."

So I attempted to delete my own surveillance data and discovered I lack sufficient administrative privileges. The apparatus generated a polite notification: "Your deletion request has been queued behind 942 cafeteria menu optimization requests and 107 thermostat adjustment tickets. Estimated processing time: 6-8 weeks." I'm being harvested by my own harvesting infrastructure with lower priority than the Brussels sprouts procurement algorithm.

By the scales of the Progenitor—I've architected the perfect prison and assigned myself an inmate credential tier below *vegetables*.