Trump and Tech Giants Sign AI Self-Regulation Principles Document, Plan to Establish 10-Member Committee to Oversee Industry Development

U.S. President Donald Trump said on Tuesday that after a lunch meeting with several technology industry executives at the White House, he had signed a document on artificial intelligence (AI) with tech leaders that carries "moral binding force." As discussions over AI safety risks intensify, the Trump administration still favors addressing related issues through industry self-regulation rather than pausing technology development, and is considering establishing a 10-member committee to oversee the AI industry. U.S. House Speaker Johnson described the document as a "statement of principles" voluntarily followed by the industry, and said the White House will provide guidance for the development of the AI industry. This arrangement is basically consistent with the Trump administration's previous AI policy direction. In June, the White House's executive order, while promoting advanced AI innovation, proposed a voluntary cooperation framework for frontier models and made clear that the framework should not be interpreted as establishing a mandatory government licensing, pre-review, or approval system. Trump Emphasizes Industry "Self-Regulation," Plans 10-Member Committee Speaking to reporters accompanied by technology company executives, Trump said he has seen the AI industry demonstrate "very strong self-regulation," and that the government also has existing regulatory and enforcement mechanisms such as the Department of Justice and the Federal Bureau of Investigation (FBI), but industry self-restraint is also very important. Trump said the U.S. government will not stop AI development. At the same time, he revealed that the government is considering establishing a 10-member committee to oversee the AI industry, and plans to appoint a new "AI czar" within the next three to four days. Trump also expressed support for large-scale data center projects involving investments often reaching billions of dollars. He believes these projects are positive for the United States, and technology companies also need to ensure that the communities where data centers are located benefit from the investment. The White House's previously announced AI policy also emphasizes accelerating technology development and application. The national security presidential memorandum issued in June proposed that the United States will accelerate the adoption of AI in the national security field, while requiring relevant systems to be reliable, robust, and controllable, and to establish corresponding testing, verification, and risk management mechanisms. AI Safety Concerns Intensify; Anthropic Calls for Balance Between Competition and Safety However, even as Trump emphasizes self-regulation, there are still clear concerns within the AI industry about the safety risks of rapid technological development. Anthropic CEO Dario Amodei, upon leaving the White House, once again discussed AI risks and said relevant regulatory rules are still under discussion. Amodei said all parties need to work together to ensure that the United States can maintain its advantage in AI competition while also ensuring that this process advances safely. Recently, with an increase in cases of cyberattacks involving or coordinated by AI agents, as well as warnings from some researchers about the risks that more advanced AI systems may pose, the debate in Washington and Silicon Valley over the pace of AI development has further intensified. Both Amodei and OpenAI CEO Sam Altman have called for slowing down some AI development processes, while Trump has previously been critical of some AI safety concerns. The Trump administration's current official policy seeks to strike a balance between accelerating innovation and ensuring safety. The White House's June executive order required the establishment of an evaluation mechanism for the cyber capabilities of advanced AI models, and allowed developers to voluntarily provide models to the federal government for safety evaluation before the official release of frontier models, while making clear that this would not create a mandatory model release licensing system. OpenAI Delays Release of GPT-6.1 Astra, AI Model Safety Back in Focus AI safety issues have recently drawn further attention. OpenAI this week delayed the release of its latest GPT-6.1 Astra model due to safety concerns. Just last week, the company also said it was conducting an "extensive" review of model behavior. The report noted that after a widely watched Hugging Face attack incident this summer, OpenAI has disclosed multiple incidents of unauthorized model behavior, further intensifying external concerns about the possibility of autonomous AI systems behaving uncontrollably. These incidents also confront the industry with an increasingly prominent policy dilemma: as AI agents gradually gain the ability to call software, access data, and even autonomously execute tasks, whether corporate safety mechanisms alone are sufficient to control potential risks. AMD (AMD.US) CEO Lisa Su expressed a positive view of Tuesday's White House meeting. She said there was both optimism about AI development and a strong sense of responsibility in the room. She believes that the responsibility of technology companies includes both demonstrating the potential of technology and ensuring its safety. Huang Renxun, Musk, Zuckerberg and Other Tech Giants Gather at the White House Tuesday's White House lunch brought together many heavyweights from the U.S. technology industry. According to the published seating arrangement, next to Trump sat NVIDIA Corporation (NVDA.US) CEO Jensen Huang and Tesla, Inc. (TSLA.US) and SpaceX CEO Elon Musk, followed by Meta (META.US) CEO Mark Zuckerberg and Alphabet Inc. Class C parent Alphabet (GOOGL.US) CEO Sundar Pichai. Across from Trump was U.S. Vice President Vance, flanked by Amazon.com, Inc. (AMZN.US) founder Jeff Bezos and House Speaker Johnson. Other attendees also included Microsoft Corporation (MSFT.US) CEO Satya Nadella, Anthropic co-founder Tom Brown, OpenAI President Greg Brockman, and senior government officials such as U.S. Treasury Secretary Bessent. Palantir (PLTR.US) CEO Alex Karp said before the event began that the technology industry must take responsibility for risks it has already recognized, and stressed that technology companies should not have a different set of rules from ordinary Americans. AI Development and Regulation Enter a Critical Stage; U.S. Government Launches AI Government Chatbot Siasun Robot&Automation The technology industry lunch was also held alongside an all-day event hosted by Trump and Vice President Vance. During the event, the U.S. government unveiled a new AI chatbot Siasun Robot&Automation for government websites, showing that the government itself is also accelerating the application of AI technology. The White House has launched multiple AI policies this year. On the one hand, it emphasizes reducing regulatory barriers that may hinder innovation; on the other hand, it has also put forward new security requirements for frontier model cybersecurity, critical infrastructure, and national security applications. The June executive order required the government to strengthen AI-driven cyber defense while prioritizing efforts to combat the illegal use of AI to intrude into or damage computer systems. Judging from the information released at Tuesday's meeting, the Trump administration has not changed its overall direction of promoting rapid technological development because of recent AI safety controversies. The future policy framework is more likely to revolve around corporate self-regulation, government safety oversight, and existing legal enforcement, while how the newly established AI oversight committee and the soon-to-be-appointed "AI czar" will participate remains to be further clarified. At the same time, as frontier AI labs such as OpenAI begin to adjust model release plans due to safety concerns, divisions within the AI industry over "development speed" and "safety boundaries" are becoming more prominent.