AI Leaders Call for Caution as Development Accelerates

In the middle of September, as artificial intelligence continued its headlong sprint forward, the people steering some of the world's most powerful AI companies began speaking openly about pumping the brakes. Dario Amodei, who runs Anthropic, one of the leading firms building large language models, went public with a straightforward message: the pace of development had become too fast, and the oversight mechanisms meant to catch problems were falling behind.

This was not a lone voice in the wilderness. Across the AI industry—among executives, researchers, and staff members—a pattern of warnings had begun to emerge. The concern was not abstract. These were people who understood the technology intimately, who had watched it evolve from theoretical possibility to practical tool in the span of a few years, and who were now flagging risks that kept them awake at night. The speed of progress, they argued, had outpaced the ability of institutions, regulators, and society itself to understand what was being built and what it might do.

The BBC decided to step away from the executive suites and the research labs and ask ordinary people what they made of all this. In Washington DC—a city where policy and technology collide, where the people who write the rules live and work—reporters went out with a simple question: Are you worried about artificial intelligence? Should it be slowed down, or is the rush to develop it justified?

What emerged from those conversations was not a unified answer. Some people saw the technology as fundamentally beneficial, a tool that would solve problems and create opportunities. Others found the warnings from industry insiders genuinely alarming. The gap between these two views was not small. It reflected a deeper uncertainty about what artificial intelligence actually is, what it might become, and whether the people building it truly understand the consequences of what they're creating.

The timing of Amodei's call for caution mattered. It came as AI systems were becoming more capable, more widely deployed, and more integrated into decisions that affected real people—hiring, lending, healthcare, criminal justice. The faster the technology moved, the less time there was to study its effects, to build safeguards, to think through what might go wrong. And yet the economic and competitive pressures pushing development forward were immense. No company wanted to be the one that slowed down while competitors raced ahead.

What the BBC found in Washington was a public still forming its views. Some residents expressed enthusiasm tempered by caution. Others were skeptical that any oversight would actually work. A few seemed genuinely frightened. Most fell somewhere in between—interested in the benefits, aware that risks existed, uncertain about what should be done. The conversation itself—the fact that it was happening at all, that people were being asked to think seriously about the technology reshaping their world—suggested that the moment of casual acceptance had passed. AI was no longer something happening in the background. It was becoming a question that demanded an answer.