{
  "id": 9195316,
  "title": "O medo real dos donos da AI: Liability",
  "url": "https://urgent.news/2026/09/22/o-medo-real-dos-donos-da-ai-liability",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-22T20:24:58.000Z",
  "source": {
    "name": "Brazil Journal",
    "slug": "brazil-journal",
    "url": "https://braziljournal.com/o-medo-real-dos-donos-da-ai-liability/"
  },
  "original_language": "pt",
  "account": "Jensen Huang, the CEO of Nvidia, dismissed predictions that humanity could be extinct by 2030 due to artificial intelligence (AI) and labeled apocalyptic statements from OpenAI and Anthropic executives as \"irresponsible.\" \"2030 will not be the end of the world. There is a 0% chance of it being the end of the world,\" Huang stated in an interview with CBS, suggesting that such claims are made by AI startups to protect themselves from legal liabilities of their autonomous systems. For Nvidia's CEO, defending a \"fledgling\" approach to AI development might be motivated by a desire to avoid potential liabilities caused by their products. \"Read between the lines,\" Huang said, \"They're not asking for more laws. They're asking to be exempt from the laws that already exist, and I think that's a problem.\" Before presenting new laws and regulations, Huang suggested applying existing ones. \"We already have all kinds of laws and rules that can be applied.\" While acknowledging the risks involved in AI development, Huang dismissed speculations about humanity's destruction as \"lacking scientific backing.\" Nvidia's CEO said the future of his business depends on the safety and reliability of his products. \"If we don't continue doing this, our value will decrease.\" Huang also noted that AI startups are transitioning from labs to product and service companies. \"Both OpenAI and Anthropic are increasing their computational capacity at an unprecedented rate – and for good reason, as their products are good and useful, which is great.\" Huang's comments align with those of Palantir founder Alex Karp, who suggested in a CNBC interview that the U.S. government could eventually \"nationalize\" major AI firms due to the liability associated with the technology's \"unlimited risks.\" \"The first line of protection is based on the principle that you are responsible for your own actions,\" Karp stated. \"So, it's a curious turn: they are asking for regulation to get rid of the first line of defense – that if you create a technology capable of destroying 10% of the planet, it entails civil and criminal liability.\" For Karp, the only way to handle such a liability might be to bring the matter to the government's attention: \"Please, audit me.\" According to Karp, what makes regulating AI difficult is that \"everyone who understands AI is on someone's payroll,\" and conflicting interests prevent mutual trust. \"In business, people often don't explicitly say what they want; they lead you to a conclusion,\" Karp commented on Amodei and Altman's cautionary speech. Even if they don't destroy the planet – or a significant part of it – AI companies are already subject to numerous types of liability. There is a flood of lawsuits brought by individuals and companies accusing AI firms of intellectual property theft. In an interview last month, Karp accused OpenAI and Anthropic of wanting to \"hook us\" in a future where their LLMs would steal their clients' intellectual property.",
  "summary": "Jensen Huang, o CEO da Nvidia, desacreditou as previsões de que a humanidade poderá ser extinta até 2030 pela inteligência artificial e chamou de “irresponsabilidade” as afirmações apocalípticas feitas por executivos da OpenAI e da Anthropic. “2030 não será o fim do mundo. Há 0% de chance de que seja o fim do mundo,” Huang […] The post O medo real dos donos da AI: Liability appeared first on…",
  "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."
}