AI pause
In plain wordsA proposed halt, temporary or indefinite, on building AI systems more powerful than today's, to give safety research and regulation time to catch up.
A proposed halt, temporary or open-ended, voluntary or legally enforced, on training or deploying AI systems beyond a given capability or compute level, intended to give alignment research, evaluation and governance time to catch up. Proposals differ widely. The March 2023 open letter Pause Giant AI ExperimentsPause Giant AI Experiments, prop. n.: A 2023 open letter, signed by tens of thousands of people including prominent technologists, that asked AI labs to halt the training of any AI more capable than GPT-4 for at least half a year. sought a halt of at least six months in training anything more capable than GPT-4GPT-4, prop. n.: OpenAI's 2023 language model, a big step up from the one behind the first ChatGPT. It scored well on many professional and academic exams and could also read images.; Eliezer Yudkowsky argued in Time the same month for an indefinite international moratorium, enforced if necessary against rogue data centers; the activist group PauseAI campaigns for an international treaty; conditional proposals would pause only when dangerous capabilities are detected, as in if-then commitmentsif-then commitment, n.: A promise by an AI developer of the form: if our model can do a particular dangerous thing, then we will have these protections in place — or we will stop.; and the October 2025 Statement on SuperintelligenceStatement on Superintelligence, prop. n.: A 2025 public statement, signed by scientists, celebrities and politicians, calling for a ban on building superintelligent AI until it is shown to be safe and the public supports it. asked for a ban on developing superintelligence until scientists broadly agree it can be made safe and the public strongly backs it. Objections include enforceability, the advantage a pause could hand to non-participating countries, and compute overhangcompute overhang, n.: A situation in which far more computing power is available than AI currently uses, so that a new method or a decision to scale up could suddenly produce a big jump in ability.: a sudden jump in capability when a pause ends.
A conditional pause only bites if the evaluations can actually detect the capability in time.
Origin
Calls to slow AI development are decades old, but the term became current with the Future of Life Institute's open letter of 22 March 2023, signed by tens of thousands of people including Yoshua Bengio, Stuart Russell and Elon Musk. PauseAI was founded in May 2023 by Joep Meindertsma.
Related
- Pause Giant AI ExperimentsPause Giant AI Experiments, prop. n.: A 2023 open letter, signed by tens of thousands of people including prominent technologists, that asked AI labs to halt the training of any AI more capable than GPT-4 for at least half a year.
- Statement on SuperintelligenceStatement on Superintelligence, prop. n.: A 2025 public statement, signed by scientists, celebrities and politicians, calling for a ban on building superintelligent AI until it is shown to be safe and the public supports it.
- compute overhangcompute overhang, n.: A situation in which far more computing power is available than AI currently uses, so that a new method or a decision to scale up could suddenly produce a big jump in ability.
- AI arms raceAI arms race, n.: A competition among countries or companies to build the most powerful AI first, in which fear of falling behind pushes everyone to move faster and cut corners on safety.
- if-then commitmentif-then commitment, n.: A promise by an AI developer of the form: if our model can do a particular dangerous thing, then we will have these protections in place — or we will stop.
- AI red linesAI red lines, n.: Limits on AI that would be forbidden worldwide no matter what — for example AI that copies itself without permission or controls nuclear weapons — which campaigners want governments to agree on.
- doomerdoomer, n.: Someone who believes advanced AI is likely to cause a catastrophe, possibly human extinction; often used as a put-down by people who think those fears are exaggerated.
Mentioned in
AI arms raceAI arms race, n.: A competition among countries or companies to build the most powerful AI first, in which fear of falling behind pushes everyone to move faster and cut corners on safety., doomerdoomer, n.: Someone who believes advanced AI is likely to cause a catastrophe, possibly human extinction; often used as a put-down by people who think those fears are exaggerated., AI red linesAI red lines, n.: Limits on AI that would be forbidden worldwide no matter what — for example AI that copies itself without permission or controls nuclear weapons — which campaigners want governments to agree on., effective accelerationismeffective accelerationism, n.: An online movement, known as e/acc, that wants AI and other technology developed as fast as possible, arguing that progress is good and that safety-driven regulation does more harm than good., Future of Life InstituteFuture of Life Institute, prop. n.: An advocacy group that warns about the dangers of advanced AI and other powerful technologies, funds research and lobbies governments. It organized the 2023 open letter calling for a pause on giant AI experiments., Machine Intelligence Research InstituteMachine Intelligence Research Institute, prop. n.: A small nonprofit in Berkeley, California, among the first to warn that smarter-than-human AI could destroy humanity, which now campaigns for a worldwide halt to building it., Max TegmarkMax Tegmark, prop. n.: A Swedish-American physicist at MIT who co-founded the Future of Life Institute, which organized the 2023 open letter calling for a pause in training the most powerful AI systems., Pause Giant AI ExperimentsPause Giant AI Experiments, prop. n.: A 2023 open letter, signed by tens of thousands of people including prominent technologists, that asked AI labs to halt the training of any AI more capable than GPT-4 for at least half a year., responsible scaling policyresponsible scaling policy, n.: A public rulebook in which an AI company promises to test each new model for dangerous abilities and to add stronger safeguards, or slow down, as its models become more capable., Butlerian JihadButlerian Jihad, prop. n.: In the Dune novels, an ancient human revolt that destroyed every thinking machine and banned them for good; in AI debates, a name for any call to stop or outlaw advanced AI., compute overhangcompute overhang, n.: A situation in which far more computing power is available than AI currently uses, so that a new method or a decision to scale up could suddenly produce a big jump in ability., deceldecel, n.: A dismissive label, used mainly by AI accelerationists, for anyone who wants AI development slowed down, paused or heavily regulated., differential technological developmentdifferential technological development, n.: A strategy of speeding up technologies that make us safer and slowing down risky ones, so that protections arrive before the dangers they guard against., if-then commitmentif-then commitment, n.: A promise by an AI developer of the form: if our model can do a particular dangerous thing, then we will have these protections in place — or we will stop., MolochMoloch, prop. n.: A name, borrowed from an ancient god of sacrifice, for situations where competition forces everyone to act against their shared interests — as when AI companies feel pushed to race ahead despite the risks., PauseAIPauseAI, prop. n.: A volunteer protest movement, begun in 2023, that pickets AI companies and lobbies governments for a worldwide treaty halting work on AI more advanced than today's best until it is proven safe., Statement on SuperintelligenceStatement on Superintelligence, prop. n.: A 2025 public statement, signed by scientists, celebrities and politicians, calling for a ban on building superintelligent AI until it is shown to be safe and the public supports it., warning shotwarning shot, n.: An AI accident or near-miss serious enough to alarm the world about AI's dangers, but not so serious that it cannot be recovered from, which could prompt action before a real catastrophe.