Views: 0
June 4, 2026, will go down in the annals of technological history. On that day, Anthropic, the company behind the Claude AI model, published a report titled "When AI Builds Itself," in which it called on the industry to consider slowing the pace of development of the most advanced artificial intelligence models.
The request doesn't come from a group of alarmist philosophers. It comes from one of the most influential companies in the sector, valued at nearly a trillion dollars. And what does it propose?
Anthropic argues for the need for a "pause button" that allows individuals, companies, and governments to take stock, strengthen oversight, and adapt governance frameworks before the capabilities of AI systems continue to advance unchecked.
It's not about shutting down AI abruptly, but about buying time for regulation to catch up with the technology. Something that, as things stand, isn't happening. And there's one statistic that explains everything. According to internal data from the report, Claude already writes more than 80% of the code integrated into its own system, and each Anthropic engineer now produces eight times more code per quarter than they did between 2021 and 2025.
Simply put: AI is already largely building itself. Humans are participating less and less in the process. And that raises an uncomfortable question: how far will this trend go?
The company warns that if AI is given the necessary processing power, it could eventually design and develop its own successor completely autonomously, a phenomenon known as recursive self-improvement. "We haven't reached that point yet, and recursive self-improvement isn't inevitable. But it could arrive sooner than most institutions are prepared for," Anthropic points out.
This is a reasonable proposal, but difficult to implement. And the problem is the context. The company itself has acknowledged how difficult it would be to get AI companies, let alone entire countries that have been locked in a race to advance their own technologies for years, to agree to a collective pause.
To make matters worse, President Trump publicly declared some time ago: "We're beating China, we're beating everybody, and I don't want to do anything that stands in the way of that advantage." This is hardly the rhetoric of someone willing to pause.
However, something is stirring in geopolitics. In a recent meeting between Trump and Xi Jinping in Beijing, artificial intelligence was on the table, with the possibility of reaching some kind of agreement to mitigate the risks this technology can pose, especially in conflict scenarios.
Among the proposed solutions is the creation of a dedicated AI hotline, with specialized teams from both countries collaborating in the event of any technological danger or emergency.
It's not much. But it's a start. And the real problem isn't technical, but rather one of collective will. There are too many economic and geopolitical interests at stake for a coordinated pause to be easy. Paradoxically, those who stand to gain the most from it are the ones with the least incentive to promote it.
Anthropic's proposal is sensible. But history teaches us that reasonable warnings are often ignored… until the problem becomes irreversible.
Let's hope this time is different, although I'm not very optimistic, and I'd love to be wrong.
EN 
ES