HALLUCINATION CLUB v1.0 — 2026 Edition

Same hallucinations. New ideas.

A year ago, the models mostly sat in a box and confidently invented citations.

Now they browse. They reason. They call tools. They write code. They operate computers. They delegate work to other agents. They remember things. They make plans. They reconsider those plans.

And, occasionally, they still make some shit up.

Welcome back to the Hallucination Club.

The 2026 edition updates the original A-OK graphic for the agentic era: more capabilities, more autonomy, more elaborate receipts—and the same fundamental uncertainty about whether the machine actually knows what it’s talking about.

IDEAS ................................ INFINITE
PROMPTS ........................ JUST ONE MORE
AGENTS .................................. A-OK
TOOLS ................................... A-OK
CONTEXT ............................ SOMETIMES
REALITY CHECKS ..................... OPTIONAL

TOTAL: STILL HALLUCINATING

Featuring our original Apes on Keys operator, now surrounded by the increasingly complicated machinery required to turn “predict the next token” into “please run my company.”

Good ideas. Questionable sources.

A-OK.