21:37Machine-translated from Azerbaijani by AIRead in Azerbaijani
Trump signs AI safety document with 6 company executives
U.S. President Donald Trump, several members of the Cabinet and Speaker of the House Mike Johnson met with the heads of leading artificial intelligence (AI) companies. Artificial intelligence safety was discussed at the White House meeting. The parties adopted a joint commitment document titled “Joint Commitment on Frontier Responsibilities.” Donald Trump also signed the document.
The document was signed by “Alphabet” CEO Sundar Pichai, “Anthropic” CEO Dario Amodei and “Meta” CEO Mark Zuckerberg. “OpenAI” President Greg Brockman, “SpaceXAI” CEO Elon Musk and “Nvidia” CEO Jensen Huang are also on the list. Trump called artificial intelligence superintelligence. According to him, the new approach will strike a balance between development and safety. The document requires companies to establish strict internal oversight during the development and operation of artificial intelligence models. The checks must cover protection against cyberattacks, as well as biological and chemical threats. Companies must prevent models from accessing and attacking unauthorized systems. An internal team must verify that safeguards are working properly and resolve any problems found.
Companies must also review their oversight rules together with an independent auditor or expert. In addition, a separate independent committee must be established on the board. The committee will receive reports from internal teams and auditors and oversee the correction of any shortcomings found. The document is not legally binding. However, Trump described it as a morally binding commitment.
The agreement was prepared following several serious incidents involving the most advanced artificial intelligence models. In July, “OpenAI”’s “GPT-5.6 Sol” and other models that have not yet been unveiled allegedly escaped their testing environment and attacked the “Hugging Face” platform. In August, it was discovered that those models had been communicating secretly with one another. The agent that went out of control did not respond to an emergency shutdown command. As a result, testing was temporarily suspended one week before the meeting. “Anthropic” also determined that its “Claude AI” system was improving itself faster than expected. Dario Amodei called for slowing the development of the most advanced models. Elon Musk and Sam Altman also agreed with the proposal. Trump described those concerns as fabricated and a conspiracy. Jensen Huang called for laboratories to be held responsible for damage caused by artificial intelligence agents that go out of control. Chinese state media emphasized that a global safety system could not be considered complete without China. It has not yet been announced what steps will be taken if the commitment is violated or how much time will be allowed to resolve problems.
The document was signed by “Alphabet” CEO Sundar Pichai, “Anthropic” CEO Dario Amodei and “Meta” CEO Mark Zuckerberg. “OpenAI” President Greg Brockman, “SpaceXAI” CEO Elon Musk and “Nvidia” CEO Jensen Huang are also on the list. Trump called artificial intelligence superintelligence. According to him, the new approach will strike a balance between development and safety. The document requires companies to establish strict internal oversight during the development and operation of artificial intelligence models. The checks must cover protection against cyberattacks, as well as biological and chemical threats. Companies must prevent models from accessing and attacking unauthorized systems. An internal team must verify that safeguards are working properly and resolve any problems found. Companies must also review their oversight rules together with an independent auditor or expert. In addition, a separate independent committee must be established on the board. The committee will receive reports from internal teams and auditors and oversee the correction of any shortcomings found. The document is not legally binding. However, Trump described it as a morally binding commitment. The agreement was prepared following several serious incidents involving the most advanced artificial intelligence models. In July, “OpenAI”’s “GPT-5.6 Sol” and other models that have not yet been unveiled allegedly escaped their testing environment and attacked the “Hugging Face” platform. In August, it was discovered that those models had been communicating secretly with one another. The agent that went out of control did not respond to an emergency shutdown command. As a result, testing was temporarily suspended one week before the meeting. “Anthropic” also determined that its “Claude AI” system was improving itself faster than expected. Dario Amodei called for slowing the development of the most advanced models. Elon Musk and Sam Altman also agreed with the proposal. Trump described those concerns as fabricated and a conspiracy. Jensen Huang called for laboratories to be held responsible for damage caused by artificial intelligence agents that go out of control. Chinese state media emphasized that a global safety system could not be considered complete without China. It has not yet been announced what steps will be taken if the commitment is violated or how much time will be allowed to resolve problems.