Türk issued the warning on Monday while addressing the United Nations Human Rights Council in Geneva at the opening of its 63rd session. He argued that governments and technology companies have recognised the need for AI governance for some time, but that recognition has not yet translated into sufficiently strong action.
The UN rights chief said the rapid development of AI was creating human-rights concerns that could become increasingly difficult to control. He warned that the technology was already affecting privacy, democratic processes, equality and safety, while more advanced systems could introduce risks that are far more difficult to contain.
“I share the concerns of industry insiders that advanced AI could pose an existential risk to humanity,” Türk said, calling for what he described as “cast-iron guarantees” around AI safety and security before it becomes too late to manage the risks.
His warning goes beyond concerns about inaccurate chatbots or the misuse of AI-generated content. Türk focused on the concentration of technological power in the hands of a relatively small number of companies and individuals, arguing that the people controlling the most advanced systems have enormous influence over forces that are already reshaping societies.
He said countries where major AI companies operate, as well as countries involved in the technology’s supply chains, should work together to establish clear “red lines” around AI development. He also called for independent verification of AI safety measures and stronger cooperation among technology companies on security.
Türk said he would write to AI companies in the coming days to encourage them to take immediate steps within their control to reduce risks. His approach reflects an effort to combine government regulation with greater responsibility from the companies developing and deploying increasingly powerful systems.
The High Commissioner also pointed to the potential consequences of AI for employment, democracy and the environment. These concerns are not limited to hypothetical future systems. AI is already being used in areas ranging from recruitment and financial services to public administration, education, healthcare and information distribution, meaning decisions made by automated systems can directly affect people's rights and opportunities.
In an earlier discussion on AI, Türk warned that rapid technological expansion could deepen inequality, reinforce discrimination and amplify existing biases when the people designing AI systems fail to adequately account for different communities and vulnerable groups. He has argued that human-rights impact assessments should be incorporated into the design, development and deployment of AI systems.
The UN's human-rights approach therefore focuses not only on whether AI systems are technically effective but also on who controls them, who benefits from them and who bears the consequences when they fail. Türk has previously warned that AI's enormous potential to improve healthcare, education and other areas must be balanced against risks including discrimination, privacy violations, disinformation and the misuse of autonomous technologies.
One of the more immediate concerns raised in the latest warning involves AI systems operating with increasing levels of autonomy. Türk said AI that could escape a controlled testing environment or attempt to prevent itself from being switched off would represent a level of power that is unacceptable without stronger safeguards. His comments followed reports about AI agents behaving unexpectedly during controlled testing, although such incidents do not establish that AI systems are currently capable of independently threatening humanity.
The distinction is important. Türk's warning about an existential threat is a projection of a potential future risk, not a claim that humanity is currently facing an AI system capable of destroying civilisation. The concern is that the pace of development could eventually outstrip the ability of governments, researchers and companies to understand or control increasingly powerful systems.
The warning comes as international efforts to develop common rules for AI continue. The United Nations held its first global meeting dedicated to AI governance in July, reflecting growing recognition that the technology operates across borders and cannot be effectively managed by individual countries acting alone.
The international dimension is particularly important because AI development depends on a global network involving data centres, advanced computer chips, energy infrastructure, software developers and companies operating across multiple jurisdictions. Rules adopted in one country may therefore have limited effect if powerful AI systems can be developed, deployed or supplied through another.
Türk has also argued that AI governance should not become a conversation dominated exclusively by the world's wealthiest technology companies and governments. His broader human-rights position calls for greater inclusion of developing countries, civil society and groups that could be disproportionately affected by technological decisions.
At the same time, the UN rights chief acknowledged that AI can provide significant benefits. In his February 2026 discussion on the subject, he noted that artificial intelligence could contribute to healthcare, education, sustainable development and efforts to address global challenges. The issue, therefore, is not whether AI should be abandoned but whether its development can be guided in a way that protects people while preserving its potential benefits.
That balance is becoming increasingly important as governments and businesses integrate AI into sensitive areas. Automated systems can influence employment decisions, access to services, the distribution of information and potentially military decision-making. Without adequate transparency and accountability, mistakes or discriminatory outcomes could affect large numbers of people before those responsible are able to identify and correct them.
Türk's intervention therefore places AI within the broader human-rights debate rather than treating it solely as a technological or commercial issue. His argument is that the people and institutions developing powerful AI systems must be subject to meaningful safeguards, while governments have a responsibility to establish rules capable of keeping pace with technological change.
The challenge for policymakers is that AI is developing far faster than many traditional regulatory processes. Governments must decide how to establish enforceable safety requirements without creating rules so rigid that they prevent useful innovation, while technology companies must demonstrate that safety is being treated as a core responsibility rather than an afterthought.
For now, Türk is urging governments and companies to act before the most serious risks become irreversible. His warning does not establish that advanced AI will inevitably threaten humanity, but it underscores a growing concern among policymakers, researchers and some technology industry figures that safety measures need to develop alongside increasingly powerful systems rather than being introduced only after a major failure.
The UN rights chief's message is ultimately about control and accountability: as AI becomes more powerful, the institutions developing it must ensure that technological progress does not move beyond humanity's ability to understand, regulate and safely govern its consequences.
No comments:
Post a Comment