Aviation Safety on Autopilot: Trump Wants AI to Write the Rules

"Flooding the zone" is how the new strategy is described at the US Department of Transportation. While the tech world argues over whether ChatGPT will replace programmers, the Trump administration has gone all-in and decided to entrust neural networks with the holy of holies: aviation safety. The idea is cynically simple: if the government machine can't keep up with the market, just press the "generate" button. This is not merely an attempt to streamline the work of clerks. It is a radical shift in how the rules that govern flying all over the world get made.

Historically, aviation safety rules have been written, quite literally, in blood. Every comma in FAA (Federal Aviation Administration) regulations appeared after thorough crash investigations, years of testing and endless rounds of approval. The process is slow and painful, but it has been proven over decades. Now there are plans to turn it into an assembly line. Officials intend to use large language models to write new rules and revise old ones, getting rid of "excessive regulation" in record time. The logic is simple: the faster the barriers come down, the faster business takes off.

The problem is that AI is neither a lawyer nor an aircraft engineer. It is a complex statistical machine that predicts the next word in a sentence. When such a machine writes code for a mobile game, the risk is bugs in the interface. When it writes operating rules for critical aircraft components, the risk falls on hundreds of people on board. The main argument of the reform's supporters is that AI can analyze thousands of pages of documentation faster than any team of lawyers. But does the algorithm understand the context of safety the way a person who has seen the consequences of technical errors does?

The move looks like an attempt at total deregulation under the guise of technological progress. Aviation has long had a term for "regulatory capture," in which the companies being overseen begin to dictate terms to their overseers. An AI trained on texts that include ones written by lobbyists for the aviation giants could turn the regulator into an automatic printer for corporate interests. Meanwhile, no official is yet ready to take responsibility for possible "hallucinations" by the algorithm in a legal text.

It is especially ironic that all this is happening amid a prolonged crisis of trust in the aviation industry. After the series of incidents involving Boeing, the world expected tighter oversight and more thorough inspections, not their automation. The new administration, however, is convinced that speed matters more than procedure. If the experiment at the Department of Transportation succeeds, we will see the same approach in every field, from medicine to environmental policy. Laws will be written faster than we can read them, let alone grasp their long-term consequences.

The main question is not even whether AI will make a mistake. It will; that is built into the nature of the technology. The question is whether a living person who can hit the brakes in time will remain in the chain. So far, "flooding the zone" looks like an attempt to tear down the old order without building a reliable replacement. We are entering an era in which flight safety may begin to depend on how good a prompt a mid-level official wrote on a Monday morning.

The key question: Is society ready for laws to be written by algorithms without human reflection?