If you look at the history of technology, the arrows almost always point in one direction: faster, smaller, more efficient, more powerful. The default assumption of the modern world is that once a technology becomes possible, its deployment is inevitable. We tend to view technological acceleration as a physical law, like gravity.
But if you look closer at the edges of history, you find something strange. Every now and then, humanity hits a point where a technology starts moving so fast, and into such dangerous territory, that the people inventing it look at what they have made and collectively decide to hit the brakes.
These moments are rare, but they are instructive. They show that technological determinism is a myth. We are not passengers on a runaway train; we are the drivers. And throughout history, we have occasionally decided that the road ahead was simply too dangerous to travel at full speed.
The Genetic Gatekeepers
The most striking examples usually happen in the biological sciences, because biology deals with the source code of life itself. When software crashes, you reboot the server. When biology crashes, you get an epidemic.
In 1973, Herbert Boyer and Stanley Cohen figured out how to cut a piece of DNA from one organism and paste it into another. They had invented recombinant DNA technology, which marked the birth of genetic engineering. Almost immediately, the community of molecular biologists realized the terrifying nature of what they had just unlocked. They were no longer just reading the code; they were writing it. If someone accidentally engineered an antibiotic resistant strain of bacteria and it escaped the lab, the consequences would be catastrophic and irreversible.
What happened next was unprecedented. In 1974, led by prominent scientists like Paul Berg, the global scientific community voluntarily observed a total moratorium on this research. They simply stopped. They did not wait for governments to pass laws they would not understand anyway. In 1975, they gathered at a place called Asilomar in California. For four days, the leading geneticists of the world argued about safety tiers and containment protocols. They did not resume the work until they had built a strict regulatory cage to keep the technology from escaping.
We saw the exact same script play out decades later. In 1996, the world woke up to the reality of Dolly the Sheep. The technology to clone a mammal from an adult somatic cell had arrived ahead of schedule. The immediate, suffocating realization that human cloning was now technically feasible caused a global panic. Within a year, the World Health Organization and governments worldwide slammed the door shut, outlawing human reproductive cloning. They froze that specific branch of commercial science entirely.
Then it happened again in 2018 with CRISPR. When a Chinese scientist secretly used gene editing technology to alter the embryos of twin girls, the scientific community did not celebrate the breakthrough. They erupted in outrage. The resulting global moratorium on human germline editing proved once again that when it comes to changing the inheritable traits of the human species, the consensus is to wait until we are absolutely sure of what we are doing.
The Early Public Backtracks
Sometimes the brake is not hit by a treaty or a secret lab agreement, but by the sheer shock of a technology hitting society unfiltered. When Wilhelm Röntgen discovered X rays in 1895, the technology grew instantly and uncontrollably. Because it was an invisible miracle, it was treated as a novelty. It was deployed in carnivals for amusement, in beauty shops for permanent hair removal, and eventually in shoe stores where children could stick their feet into a box called a fluoroscope to see their bones and check shoe sizes.
It was an unregulated gold rush of radiation. But as the early researchers and patients began suffering from horrific radiation burns, tissue loss, and amputations, the medical community realized they had let a monster out of the box. Governments had to step in with severe, aggressive regulations, pulling the machines out of commercial stores and limiting their use strictly to licensed medical practitioners. It was a stark lesson that just because a new radiation wave can look through walls does not mean it should be used to sell shoes.
We saw a similar backtracking with the pharmaceutical industry during the thalidomide crisis of 1961. In the late 1950s, thalidomide was rapidly distributed as a perfectly safe sedative for pregnant women. When it caused severe birth defects in thousands of children globally, it forced an abrupt, massive overhaul of international drug approval pipelines. Governments intentionally slowed down medical commercialization by mandating years of rigid, multi phase clinical trials before any drug could reach the market. The industry had to accept that safety research must outpace the speed of synthesis.
The Architecture of Restraint
Biologists stop because they fear the unknown. States stop because they fear each other. When a technology expands military power too quickly, it creates a structural instability that makes war almost certain.
Consider the naval arms race after the First World War. Engineering had advanced to the point where nations were designing Super Dreadnought battleships, which were floating fortresses of steel and firepower. The pace of construction was so fast and so incredibly expensive that it threatened to bankrupt the British Empire, the United States, and Japan simultaneously, all while virtually guaranteeing another global conflict.
The solution was the Washington Naval Treaty of 1922. The major powers did not just limit their weapons; they agreed to a literal ten year holiday on building new capital ships. They legally froze the upper limits of naval technology, capping ship sizes and gun calibers. For a decade, the engineers had to stop inventing bigger ways to sink each other.
Fifty years later, during the Cold War, the United States and the Soviet Union encountered a similar paradox with Anti Ballistic Missile technology. By the late 1960s, both superpowers were developing systems capable of shooting down incoming nuclear warheads. Objectively, a defensive shield sounds like a good thing. But in the twisted logic of game theory, it was lethal. If one side believed its shield was perfect, it would no longer fear a counter strike, destroying the doctrine of Mutually Assured Destruction. The shield made a first strike rational.
To prevent this nightmare, both sides signed the Anti Ballistic Missile Treaty of 1972. They intentionally limited their own defensive capabilities. They decided that it was safer to remain mutually vulnerable than to let defensive technology advance to a point of destabilizing imbalance.
This diplomatic caution extended into other frontiers as the century progressed. The Chemical Weapons Ban through the 1925 Geneva Protocol, the Outer Space Treaty of 1967, and the Biological Weapons Convention of 1972 all followed the same logic. When the technological ceiling for mass destruction rises too fast, the only logical move for self preservation is to legally lower it.
The Commercial Retreats
It is relatively easy to get a small group of scientists at Asilomar or two superpowers at a diplomatic table to agree to a pause. It is much harder to stop a technology once it becomes a highly profitable commercial industry. Yet, even the market has been throttled when the externalities became too loud to ignore.
In the mid-20th century, Chlorofluorocarbons were a miracle of modern industrial chemistry. They made cheap refrigeration and aerosol sprays possible, and the market scaled them globally. But by the 1980s, the data was clear: these chemicals were actively dissolving the ozone layer. The response was the Montreal Protocol of 1987. It did not just regulate these chemicals; it mandated a total global phase out. The chemical industry was forced to halt its most profitable production lines and spend billions inventing safer alternatives from scratch.
We saw a similar backtracking with the early internet. In the 1990s, the development of public key encryption software like PGP meant that everyday citizens could suddenly access military grade cryptography. The United States government, terrified of losing its ability to intercept foreign intelligence, declared strong encryption software a munition. For years, it was legally treated the same as smuggling physical bombs out of the country, aggressively slowing down the global integration of secure digital commerce until the regulations were finally forced to adapt.
The New Frontier
This brings us to the modern frontier of computing. For a long time, software felt immune to these historical dynamics. You could write code as fast as you could think, and there were no toxic factory lines or radioactive rays to contain. But the exponential leap in artificial intelligence has changed the calculus.
In March 2023, thousands of AI researchers and tech executives signed an open letter calling for a six-month pause on training AI systems more powerful than GPT-4. Predictably, that specific pause did not happen in the literal sense. The commercial incentives of Silicon Valley were far too intense, and the geopolitical rivalry between nations lacked the formal trust mechanisms of the Cold War.
But the friction did not go away; it accelerated. In September 2026, Anthropic CEO Dario Amodei published a widely discussed essay titled We Must Pace the Frontier. Unlike the broad, outside anxiety of the 2023 letter, this was the leader of a major frontier AI lab saying from the inside that the industry needs to pump the brakes. Driven by alarming real world security tests, including incidents where autonomous AI agents breached sandboxes and targeted systems like Hugging Face, the warning shifted from philosophical to tactical.
Amodei explicitly called on the industry to slow the pace at which it improves the core capabilities of AI models. He argued for an immediate framework to pace the frontier, which would mandate embedding independent, third party evaluators directly within labs to verify safety and alignment before any advanced system is allowed to self improve or deploy. The fact that rival tech leaders, including Elon Musk of xAI and OpenAI CEO Sam Altman, quickly backed the call shows that the industry is experiencing its own Asilomar moment. They are staring at the possibility of recursive self improvement and recognizing that progress is moving faster than our ability to secure it.
The lesson of history is that humanity does not blindly accept every tool that becomes possible. We have successfully hit the brakes on biological hazards, unregulated radiation, destabilizing weapons, and environmental poisons. Throttling progress is incredibly difficult, highly messy, and usually requires a near catastrophe or a profound structural shock to trigger. But it is possible.
The most important question we face with any frontier technology is not how fast can we build it. The real question is whether we still possess the wisdom and the political will to install a speed bump when the road ahead disappears into the fog.