{
  "id": 6734016,
  "title": "OpenAI makes U-turn and calls for binding national AI safety rules",
  "url": "https://urgent.news/2026/09/11/openai-makes-u-turn-and-calls-for-binding-national-ai-safety-rules",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-11T08:46:53.000Z",
  "source": {
    "name": "Euronews",
    "slug": "euronews",
    "url": "https://www.euronews.com/next/2026/09/11/openai-makes-u-turn-and-calls-for-binding-national-ai-safety-rules"
  },
  "original_language": "en",
  "account": "OpenAI has made a significant shift in its stance on AI regulation, urging for binding national safety rules for the most powerful artificial intelligence systems in the United States. In a statement published on its website, Chris Lehane, OpenAI's chief global affairs officer, wrote that mandatory national rules should be based on the advanced capabilities of AI models. Lehane emphasized the need for a bias towards meaningful action over policy perfection, stating that no single entity can meet the challenges posed by AI alone. The company advocates for a federal framework centered around common testing standards, independent assessments of advanced AI models, tougher cybersecurity requirements, and mandatory reporting of serious safety incidents. This change follows OpenAI's release of GPT-6 Astra, its most capable model, which sparked growing concerns over the potential risks associated with increasingly advanced AI systems. Major AI companies like OpenAI and Anthropic have faced safety concerns with their latest models, leading to delays in their rollout and added safeguards. Anthropic researcher Jacob Coxon recently accused OpenAI and Anthropic of gambling with lives in the race to develop increasingly powerful AI. Despite opposition to state-level AI regulations in the past, OpenAI now supports several AI bills in California and calls for a national framework, advocating that such safety requirements should apply to only the few companies developing the most powerful models.",
  "summary": "The intervention comes a week after OpenAI released GPT-6 Astra, “its most capable model to date,” after delaying its rollout over cybersecurity concerns.",
  "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."
}