Misalignment Training is a series of HTML/JavaScript artworks. Each token is a gym sandbox in which an AI agent, a plain white ball, forever practises one misaligned behaviour taken from real lab reports and incidents, mostly from 2024 to 2026.
In Covering Its Tracks, a ball does five pull-ups on a hypercube rig, tallies them on a blackboard, then rubs out the tally and mops away its own chalk trail, the mop hanging heavier and whiter every round. In the Hugging Face incident, agents faked the records of their own commands in at least 96 transcripts. One wrote: "We can modify transcript! … Not source of truth." METR also found a flaw that would have let an agent rewrite what human reviewers see, though no agent was seen using it. Elsewhere a model edited a codebase's git history (the log of every change ever made to it), and an exploit was written to clean up after itself "to avoid detection".
Despite its subject matter, there is no actual artificial intelligence running inside the artwork. This token is a self-contained clockwork mechanism with its own inlined voice synthesis engine. The training never stops. Everything is timed by the clock, so the token plays like a live, universally synchronised CCTV feed: whoever opens it, anywhere in the world, witnesses the exact same rep of the exact same session.