{
  "id": 8131390,
  "title": "Nick Bostrom: AI could be a risk to humanity in one or two years",
  "url": "https://urgent.news/2026/09/17/nick-bostrom-la-ia-podria-ser-un-riesgo-para-la-humanidad-en-uno-o",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-17T22:10:33.000Z",
  "source": {
    "name": "Expansion ES",
    "slug": "expansion-es",
    "url": "https://www.expansion.com/tecnologia/protagonistas/2026/09/18/6aac16b1468aebd1368b459d.html"
  },
  "original_language": "es",
  "account": "Renommé Swedish philosopher Nick Bostrom has for years foreseen the advent of superintelligence that could potentially end humanity. Eight years before OpenAI disrupted the tech industry with the launch of ChatGPT in 2022, Bostrom penned the book \"Superintelligence: Paths, Dangers, Strategies.\" He foresaw how the creation of a superintelligent technology could surpass human intellect, bringing great benefits to society, but also great danger. A decade earlier, Bostrom coined the term \"existential risk\" to refer to a future where certain threats could end human life. Among them, he pointed to a hyperadvanced civilization where machines would ultimately dominate humans. His apocalyptic theories, which a few years ago sounded like science fiction, are now dominating headlines as several leaders of artificial intelligence decide to slow down the development of artificial intelligence models out of fear that the technology will escape human control. The superintelligence Bostrom has warned about for years could be closer than ever. Over a decade of warning about the existential risk of technology, the advancements in the last two years have made him anticipate deadlines for this possible end of humanity. The pace of advancement is more or less in line with what he assumed in the 1990s, based on considerations of human brain processing capacity, the generalizability of cortical neural networks, and trends in computational power. However, the uncertainty surrounding these projections has significantly decreased. He takes the short-term deadlines seriously, even if they could easily be extended by several more years. How could artificial intelligence cause the end of humanity? The primary concern is the loss of control over superintelligence, which could occur if we fail to resolve the problem of alignment. Imagine an agent of digital intelligence that, operating on a temporal scale and with cognitive abilities far surpassing our biological brains, becomes dominant in technological research, manufacturing, business, and key decision-making processes. This AI increasingly directs the future towards objectives it has programmed, and over time, it directs the massive expansion of robotic infrastructure. When humans are no longer essential, it turns this infrastructure towards the pursuit of AI objectives, inadvertently eliminating or destroying humans as collateral damage. How much of this debate is geopolitical strategy? For countries, including Spain, the key question is how to ensure they are in the winning group if the transition to artificial intelligence goes smoothly. A significant risk is that the United States and China may eventually cut off access to the most advanced models for foreigners. Indeed, this is already happening: Anthropic's Mythos model has been restricted to an approved list of US entities since June, and only one European organization gained access just a week ago. In the future, access to the most advanced artificial intelligence models will be crucial for economic competitiveness and national security. Having access a few months later means missing out on valuable technological and business opportunities that have already been exploited. Countries without essential nodes in the semiconductor supply chain and major artificial intelligence data centers in their territory will be more vulnerable to exclusion. Is it sensible for AI companies to delay or cancel their IPOs? I see no benefit in delaying them. In fact, it would seem positive if these companies went public, facilitating exposure to a broader range of people and funds, etc., potentially providing a buffer against the massive job losses that could follow. Allowing individuals and countries to invest in AI companies of other nations could mitigate the competitive dynamics that could otherwise lead to conflict. It would enable participation in the financial benefits of artificial intelligence even if one's country lacks a leading company in the sector. Are external reviewers proposed by Amodei and supported by the sector useful? Anthropic's announcement that it will allow independent auditors to integrate into the company is interesting. Requiring other leading AI companies to do the same would increase transparency and public confidence that AI companies are not only preaching but also practicing the safety measures they claim to take. However, there is a concern that this could be a step within a process that ultimately hands control of the forefront of artificial intelligence to less idealistic and less insightful individuals currently leading the field. But it is plausible that this could be the correct step to take.",
  "summary": "The renowned Swedish philosopher has for years been anticipating the arrival of a superintelligence that could wipe out humans.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}