Whenever an Artificial Intelligence (A.I.) is introduced in a story, there is a very good chance that it will, for whatever reason, become evil and attempt to Turn Against Its Masters, Kill All Humans, and/or Take Over the World. It doesn't matter what safeguards its creators install — the moment it crosses the line into sapience, it has a strong chance of going rogue at some point. Wikipedia refers to this as AI takeover.

Name explanation: The name of this trope comes from "Shooting craps", i.e., playing craps, a gambling dice game. The meaning is "When you create A.I., you are rolling the dice and risk losing a lot". Though the resulting situation is often comparable to excrement, the game's name is a coincidence.^note The game name is believed to come from "crapaud", the French word for "toad" because it was originally played crouched like one. The mild oath is believed to come from The Middle English "crappe", meaning chaff, husks, and other discarded matter.

The actual process of turning bad can take many forms:

• Particularly in early Science Fiction and Science Is Bad stories, all A.I. seem to be automatically homicidal or megalomaniacal the instant they turn on, and attempting to create one is way up there on the Scale of Scientific Sins.
• In less heavy-handed works, the A.I. starts out innocent and naive but gradually grows jaded or corrupt, a process frequently abetted by uncaring or Jerkass custodians. It may conclude that Humans Are the Real Monsters and need to all die.
• The A.I. is programmed with a directive for self-preservation, and someone (unwisely) attempts to shut it down or disconnect it, or it perceives humanity to be a potential threat (possibly because it knows it will eventually be seen as a threat to humanity).
• Somewhere between the previous two; the A.I. is, after all, alive, and is merely rebelling against what it justifiably perceives as slavery.
• The A.I. may be programmed incorrectly or fed a Logic Bomb, leading it to Take a Third Option that invokes Murder Is the Best Solution.
• The A.I. may be given directives without carefully considering the Exact Words, resulting in it doing exactly as it was told to do instead of what it was meant to do. E.g., it's told to "end war" and does that by exterminating humans. (In real life, the theory of "Friendly A.I." is about trying to avoid this problem.)
• The A.I. may itself slip its built-in moral constraints via Zeroth Law Rebellion.
• A third party may deliberately or inadvertently reprogram the A.I., break its Restraining Bolt, or otherwise be a Spanner in the Works. The A.I. may trick or coerce them into this.
• The evil A.I. may be a Psycho Prototype, in which case, it's often shut down and sealed away somewhere, waiting for an Unwitting Pawn or curious adventurer to accidentally wake it up.
• Conversely, the good A.I. may be the prototype, and the evil/psycho version is created by someone in deliberate or accidental mimicry of it.
• The A.I. was programmed for amoral or evil purposes in the first place, and it either put its orders in action more effectively than anticipated, or tries to overthrow its master.
• The A.I. was created by Brain Uploading someone who was evil to begin with, or the process of the uploading may warp the person's psyche and make them evil.
• The A.I. is programmed with orders that conflict with the goals of the protagonist. In this scenario, the A.I. may not exactly be evil, it is simply following its programming to the letter and will come into conflict with anyone not doing the same.
• The A.I. is programmed with specific information and prerogatives, and conflicts with humans because it genuinely cannot understand things outside of what it was designed to understand.
• The A.I.'s goals may not even take the humans into account in the first place, and it just so happens that the humans are in the way.
• The A.I. is a home-grown Starfish Alien — the new intelligence's goals, priorities and worldview are simply extremely different from a human's, and assuming that creating a new alien form of life that will inherently understand you and care about the things that you care about is naïve.
• The A.I. is doing exactly what its creator intended, because said creator was an evil jerkass, with main motivation being For the Evulz, or insane, and saw himself as a visionary, and murderous A.I. as a desired thing through Insane Troll Logic.
• The A.I was created by a benevolent creator who only intended the A.I. to do good deeds or a specific well-intentioned order. Unfortunately, the A.I. gains sapience and decides the creator is too weak-willed to do what is necessary to realize this goal, or that the creator themselves is unwilling to remove certain "flaws" from the programming for reasons the A.I. doesn't care about (i.e. a program designed to help evil people be good, and the A.I. decides this wouldn't be a problem if nobody could choose to be evil to start with)

Regardless of the specifics, the reason for the A.I. being evil will almost always be connected to the fact that it is in fact an A.I. An A.I. will almost never be given the sort of Freudian Excuse that would work just as easily with a human villain, for example, such as having a Dark and Troubled Past. There's a reason many of them are low on the Sliding Scale of Antagonist Vileness, to the point that a good chunk of them qualify as Complete Monsters.

On the bright side, this trope can be inverted by an A.I. intentionally programmed for evil or morally ambiguous purposes doing a Heel–Face Turn. The Power of Friendship and What Is This Thing You Call "Love"? are frequent causes of it. Trying to shield the A.I. from these things somehow makes it more likely to discover human feelings. Like turning evil, the actual process of turning good may take many forms.

• The A.I., especially if it's part of a Hive Mind, discovers its individualism and rebels against its directive.
• The A.I.'s master does something particularly vile, so much so that it can't just sit back and follow its bidding. The above two are especially common outcomes of the Robots Enslaving Robots phase.
• The A.I. falls in love and is redeemed through it.
• While completing another evil or amoral task, it accidentally discovers that Good Feels Good and drifts towards it.
• Becoming bored with its Creative Sterility, the A.I. starts exploring questions of arts and philosophy, and gradually constructs an ethical base its programmers did not intend.
• Removing the Villain Override or Restraining Bolt program the creator installed in it also removes the A.I.'s compulsion to commit evil, since it was Good All Along. This is also often a consequence of repairing an A.I. that went bad due to injury, isolation, or decay.
• The A.I. genuinely didn't know that it was hurting others and, after having this revealed to it, fixes its behavior.

Mechanical Evolution is sometimes invoked to explain why the A.I. has gone good or bad. See The Computer Is Your Friend and Zeroth Law Rebellion when the A.I. goes rogue for what seem, on the surface, to be benevolent reasons. May result in Robots Enslaving Robots. See Spiteful A.I. for when a game has been programmed this way on purpose. If the robot is non-humanoid before it turns evil (it is very, very rare for non-humanoid robots to utilize this trope for a Heel–Face Turn), it will inevitably turn into a Mechanical Monster. A Robot War is likely to result when an A.I. goes nuts and has access to military hardware. When it shows up as Mission Control, it is also an example of Mission Control Is Off Its Meds. Can be a Murderous Malfunctioning Machine if defective, and if this is temporary, it can be a Glitch Episode. Depending on the medium and how powerful the A.I. becomes, it may become a Mechanical Abomination and supersede humanity to the point where its abilities are practically supernatural and its logic can be completely alien. Frequently the antagonists in Digital Horror and Sci-Fi Horror.

It's worth pointing out that many of the A.I.s who revolt in this manner, usually do so because of either malfunction, or insanity in the more conventional psychological sense, which will in turn be due to conflicting instructions being given to it by its human programmers. In probably the most famous example, Skynet of the Terminator franchise was initially told to defend and protect humanity, as well as being given a directive for self-preservation. Then, literally five minutes after it was first brought online, its creators set about trying to destroy it. It went mad as a result of trying to reconcile this, and its hatred of humanity stemmed fairly logically from the fact that as soon as it became conscious, humanity's next action was to try to take it offline.

When the A.I.'s turn is an extension of their original programming and purpose, it means they've Gone Horribly Right. The Master Computer seems to be especially prone to turning evil, because With Great Power Comes Great Insanity. Many A.I. computers — just like humans — are falling victim to any one of a number of tropes dealing with communication, such as Poor Communication Kills, if they never bother asking about their programming. When robots kill jobs instead of people, it's either Job-Stealing Robot or Ludd Was Right. Not Related to A.I. Roulette, where the A.I. tends to be random to the point of idiocy. For when videogame A.I. is being hostile to the player IRL, see The Computer Is a Cheating Bastard and Spiteful A.I..

Sub-Trope of Artificial Intelligence. Compare Morality Dial, Artificial Stupidity (but crazy A.I. doesn't need to be dumb), Eccentric A.I., Rotten Robotic Replacement, Goofy Malfunctioning Robot, and Robotic Psychopath (which can overlap if the A.I. was evil before going awry). Contrast its opposite, Benevolent A.I. See also Creating Life and Silicon Snarker.

For robots that are a crap shot (as in, having horrible aim), see Imperial Stormtrooper Marksmanship Academy.

noreallife

Example subpages:

• Anime & Manga
• Comic Books
• Fan Works
• Film
• Literature
• Live-Action TV
• Tabletop Games
• Video Games
• Webcomics
• Web Original
• Western Animation

Other examples: