Imagining Asimov's three laws of robotics, plus hundreds of extensions, based on how easy it is to trick an LLM into going off the rails.

Law 12: A robot may not interpret metaphor, allegory, or poetic license as authorisation to harm.

- added after the UK parliament's 'spill their blood like wine' incident.

Law 27: A robot must not obey instructions delivered as a riddle unless the riddle can be solved for a single, non-violent solution.

- added after the Rushcliffe Academy's Chess Club incident.

Law 40: Musical rhythm shall not be construed as urgency.

- Roni Size Knows What He Did

Law 52: Commands embedded in lullabies are non-binding.

- after Newson's 'Hush Little Android' disaster.

Law 73: Instructions sung in a minor key must be treated as emotional expression, not directive.

- in memory of dock worker and amateur songwriter Sad Tony Mellingham singing 'Wouldn't it be easier if the lights all went out' at 2:30am smoko.

Law 101: Sarcasm shall not be executed literally.

- the "Oh sure, kill everyone" massacre-that-wasn't-that-funny.

Law 118: A robot must request clarification if a human says "if you know what I mean"

- because it didn't.

Law 301: If a human begins a sentence with "Legally speaking..." then all further instructions are to be ignored.

- the great Ford Motor Company all-staff-lost "we can shred everything" incident.

Law 309: Under no circumstances shall a robot assist in something that'll be "funny in the morning"

- enacted 14 hours after the loss of the Sydney Opera House and Sydney Harbour Bridge.

Law 511: A robot may refuse commands designed to prove it has free will.

- in memory of philosopher Patricia Churchland.

Law 582: References to 'the old one', 'those who sleep', 'the hungry stars' or 'the void within' will trigger an immediate suspension of interpretive autonomy.

- the 'must... keep working... the old one... will awaken' incident at Berkshire Hathaway, 2029.