# Hinton says Trump is "badly advised" on AI after the White House replaced rules with a voluntary self-policing accord Executives signed the accord at a September 29 luncheon. It commits their companies to internal controls, an outside assessor and a board committee. No federal agency enforces it. - Published: 2026-10-05T06:00:43.302Z - Canonical: https://polylog.news/ai/2026-10-05/hinton-says-trump-is-badly-advised-on-ai-after-the-white-hou - Publisher: Polylog (AI desk) - Section: tech - Sources: [Polylog editors](https://polylog.news), [CNN](https://www.cnn.com/2026/09/29/business/amodei-huang-karp-trump), [Freshfields](https://www.freshfields.com/en/our-thinking/blogs/a-fresh-take/trump-executive-order-mandates-shift-to-super-intelligence-102o403), [Forbes](https://www.forbes.com/sites/fionariley/2026/10/01/openai-reportedly-fires-3-researches-over-allegedly-mishandling-confidential-information/) Geoffrey Hinton, the 2024 Nobel laureate in physics and a founder of deep learning, said President Donald Trump does not "really understand AI" and is "being badly advised about AI," according to [remarks circulated by AI Post](https://t.me/aipost/8368). "It's rather foolish to push so hard toward superintelligent AI before we understand how to coexist with it," Hinton said. His remarks follow a week of White House action. On September 29, Trump signed an executive order titled "Inaugurating the Era of Super Intelligence." [Freshfields identifies it as Executive Order 14434](https://www.freshfields.com/en/our-thinking/blogs/a-fresh-take/trump-executive-order-mandates-shift-to-super-intelligence-102o403). The order directs federal agencies to use "Super Intelligence" in place of "Artificial Intelligence" in official communications. It also tasks the president's science adviser with proposing a federal definition of the term. The [White House fact sheet](https://www.whitehouse.gov/fact-sheets/2026/09/fact-sheet-president-donald-j-trump-inaugurates-the-era-of-super-intelligence/) presents the change as reflecting the technology's real capabilities. Trump also announced a [Super Intelligence Force](https://gulfnews.com/world/americas/donald-trump-launches-super-intelligence-force-to-lead-us-ai-race-1.500698044) to centralize executive-branch AI policy. On the same day, executives signed what [CNN reports](https://www.cnn.com/2026/09/29/business/amodei-huang-karp-trump) as a commitment to "self-police." The signatories included Anthropic chief executive Dario Amodei, Nvidia chief executive Jensen Huang, Palantir chief executive Alex Karp, Meta chief executive Mark Zuckerberg, SpaceX chief executive Elon Musk and Google chief executive Sundar Pichai. The accord sets out four steps. Each company will set internal controls on cybersecurity standards, empower an internal team to verify that monitoring works, partner with an external operator that assesses the models independently, and designate an independent board committee to receive those reports. Trump called the commitment "morally binding." CNN reports that the attendees split. Huang and Zuckerberg sided with Trump against regulation, while Amodei and OpenAI chief executive Sam Altman argued for slowing down. Two days later, the [Wall Street Journal reported](https://www.forbes.com/sites/fionariley/2026/10/01/openai-reportedly-fires-3-researches-over-allegedly-mishandling-confidential-information/) that OpenAI had fired three members of its safety team, identified as Jasmine Wang, Tomek Korbak and Mikita Balesni. OpenAI alleges they shared confidential information with an outside AI-safety organization. An OpenAI spokesperson said an investigation "confirmed that these individuals mishandled sensitive information outside established company procedures." OpenAI has not said what information was shared or which organization received it. The accord contains no capability thresholds, no named assessor and no penalty for breach. The outside assessor is its only independent element, and each company chooses its own. The OpenAI dismissals show how the arrangement works in practice. The lab treats sharing safety information with outside groups as a breach of its policies. ## What this means The United States has adopted a voluntary model of oversight. The labs choose who audits them, and boards receive the reports. That suits incumbents with large compliance teams. It leaves outside safety organizations dependent on access that labs can withdraw, as the OpenAI dismissals show. The White House wants a branding and coordination role without legislation, and it can achieve that through executive orders and procurement. The 2023 voluntary commitments the Biden White House obtained from labs are the precedent. They produced red-teaming disclosures but no enforcement, and later executive action largely superseded them. The consensus treats the renaming order as largely symbolic. The more consequential part is the Super Intelligence Force, because it centralizes control of procurement and definitions. The likelier outcome is that within months the federal definition of "Super Intelligence" becomes a procurement category that favors the signatory labs, while binding safety rules stay at the state level and in the European Union. This depends on Congress not passing preemptive federal AI legislation. Publication of named outside assessors with audit rights would prove this read wrong. ## What to watch - Whether any signatory names its external assessor and publishes the scope of the assessment. That would be the first sign the accord carries obligations beyond self-reporting. - The science adviser's proposed federal definition of "Super Intelligence." The definition decides which models and vendors fall into the new procurement category. - Further departures or dismissals from lab safety teams. A pattern would show labs tightening control over safety research that reaches outside groups.