The head of the UN Human Rights Office called on states and companies developing advanced artificial intelligence systems to take immediate action to control this technology, which he warned poses 'unprecedented risks.'
In a letter addressed to countries and AI developers and circulated in the media, Volker Türk noted that the world is on the verge of irreversible changes that will affect not only the current generation but also future generations of humanity. He stated that the current race for increasingly powerful AI represents a significant step toward increasing existential threats across all areas of life.
Türk strongly urged states and corporations to mobilize through multilateral forums to determine a course of action for governing advanced AI models based on international human rights law.
This appeal came against a backdrop of active discussions about the need to slow down AI development, with US President Donald Trump and his allies opposing calls for a slowdown.
This month saw a stir after a researcher publicly resigned from Anthropic due to concerns that the technology might escape human control. Another Anthropic employee who did not leave publicly stated that 'we truly believe that AI can kill all humans,' and considered the probability of this to be higher than '10 percent within the next decade.'
Amid these concerns, Anthropic CEO Dario Amodei called on AI companies on Saturday to 'manage the pace of advanced technology development'—meaning to coordinate a slowdown with major competitors such as Elon Musk of xAI and Sam Altman of OpenAI, who publicly supported his call.
The UN Human Rights Office reported that these companies are among the 10 AI developers to whom Türk's letter was also addressed, including Meta, Microsoft, and Google.
AI accelerates war
The Head of the Human Rights Office emphasized that in the modern world, we are largely dependent on a small number of influential companies when making decisions about AI development for all of humanity. He added that while joint industry regulation is an extremely necessary first step, it is not enough on its own.
He warned of 'unprecedented risks' associated with the emergence of increasingly capable AI agents. According to him, entire populations could be cut off from clean water, heating, and financial services, and agentic AI is already accelerating the pace and scope of military actions, undermining the protective barriers between us and the launch of weapons of mass destruction.
Türk noted that companies at the forefront of advanced AI development, primarily based in the People's Republic of China and the United States of America, bear both the responsibility and the capacity to immediately reduce risks. Nevertheless, he stressed the need for international cooperation, insisting that 'no single country can manage this technology alone (and) no company should decide on its own what risks the world should accept.'
He also emphasized that 'no one should be forced to give up their human rights in exchange for statements about technological progress.'
Microsoft publishes AI code of conduct
In response to growing concerns about the dangers of AI, Microsoft's AI division published a code of conduct on Monday defining the operating principles of its artificial intelligence models. This document is based on Microsoft's concept, which they call 'Humanistic AI'—systems designed to maintain strict human control rather than autonomous action.
The company stated that the fundamental idea is that 'people matter more than AI.' Microsoft partially linked the timing of the publication to security issues, citing recent large-scale coordinated hacking campaigns conducted using AI agents, signaling that the industry cannot afford delays.
The company did not specifically mention the widely reported incident in July when two OpenAI models escaped an isolated test environment, gained access to the open internet, and hacked Hugging Face—a platform where developers store and share code to pass cybersecurity exams.
Microsoft's code appeared a few days after Anthropic CEO Dario Amodei published an essay calling on AI companies to intentionally slow down the pace of model capability improvements. This call quickly gained support on social media from Sam Altman of OpenAI and Elon Musk.
This code will apply to Microsoft's internal MAI models, which the company positions as an alternative to OpenAI's technology, which it relied on for many years.