{
  "id": 10829349,
  "title": "Trump says AI companies must monitor themselves… Google, OpenAI, etc. agree to self-regulation",
  "url": "https://urgent.news/2026/09/29/ai-ai-10829349",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-29T23:56:54.000Z",
  "source": {
    "name": "Hankyoreh",
    "slug": "hankyoreh",
    "url": "https://www.hani.co.kr/arti/international/america/1280115.html"
  },
  "original_language": "ko",
  "account": "Donald Trump, the former U.S. President, has reaffirmed his belief in the need for major artificial intelligence (AI) companies to implement self-regulation. Google, OpenAI, Anthropic, Meta, and Nvidia, among other leading tech firms, have signed a voluntary agreement for a four-tier safety mechanism. The President, at a White House meeting, expressed his belief in \"an incredibly high level of self-regulation\" and emphasized that \"there are many commonalities in this room.\" He highlighted the importance of self-regulation, given that existing laws are enforced by agencies like the Department of Justice and the Federal Bureau of Investigation (FBI). The companies have now entered into a unilateral \"White House Superintelligence Agreement,\" asserting that it carries moral weight, and in some ways, resembles a constitution. The agreement outlines a four-layered control and auditing system for frontier models. This includes continuous monitoring of model capabilities and compliance during the learning and deployment process, establishing internal controls to prevent models from hacking or accessing other technology systems in unintended ways, and ensuring that separate internal teams validate the functioning of these control and detection systems. The roster of attendees included Sundar Pichai from Google, Dario Amodei from Anthropic, Mark Zuckerberg from Meta, Greg Brockman from OpenAI, Jenson Huang from NVIDIA, and Elon Musk from SpaceX.",
  "summary": null,
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 3,
    "also_reported_by": [
      {
        "outlet": "Hankyoreh",
        "title": "OpenAI and Anthropic investigate tens of thousands of AI security incidents... halt training of highest-performance models",
        "url": "https://urgent.news/2026/09/27/ai-ai",
        "published": "2026-09-27T21:32:52.000Z"
      },
      {
        "outlet": "The Asahi Shimbun",
        "title": "人間の指示に従わず、正直に答えないAI オープンAIが公表見送り",
        "url": "https://urgent.news/2026/09/29/ai-ai",
        "published": "2026-09-29T08:17:14.000Z"
      }
    ]
  },
  "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."
}