U.N. human rights chief Volker Türk called on governments to agree on “red lines” for artificial intelligence and establish independent verification of AI safety, as countries bring competing proposals for governing the technology to the Human Rights Council.
Türk said on Monday he would write to AI companies in the coming days urging them to take immediate steps within their control to reduce risks from increasingly powerful systems, while calling for governments that host AI companies and those involved in their supply chains to agree on common limits.
“We need independent verification, and much stronger collaboration on security within the industry,” Türk told the opening of the council’s 63rd session in Geneva.
The intervention comes as artificial intelligence takes a larger place on the council’s agenda but governments have yet to agree on a single framework setting out how the technology should be developed and governed in accordance with human rights.
Bangladesh, Egypt and Nepal are leading a new resolution on the human rights implications of AI during the current session, while Armenia is introducing a separate measure addressing women and girls in the context of digital technologies. South Korea has pursued another council initiative on new and emerging digital technologies since 2019.
The proliferation of initiatives reflects growing interest among governments in shaping AI rules, but also raises questions about whether the different efforts will converge around common standards.
“The need for AI governance is widely acknowledged – but where is the action?” Türk said. “Delay only benefits the massive tech companies, their owners and enablers.”
Türk also warned about the concentration of control over the technology.
“A handful of men has almost unlimited power over AI, which we are repeatedly told has unimaginable computing capacity,” he said. “They talk about freedom, but on closer inspection, this turns out to be little more than the freedom to exploit our data.”
AI governance is already developing through several parts of the international system. U.N. member states have established a Global Dialogue on AI Governance and an Independent International Scientific Panel on AI, while governments and regional organizations are developing their own regulatory approaches.
Türk welcomed those developments but said stronger safeguards were needed as the capabilities of advanced AI systems increase.
“I share the concerns of industry insiders that advanced AI could pose an existential risk to humanity,” he said. “AI that escapes its testing environment, or blackmails developers to prevent itself from being turned off, is AI that is too powerful.”
Türk called for an “all-out effort” to establish what he described as “cast iron guarantees” around AI safety and security before the technology advances further.
His proposed red lines were not defined in Monday’s speech, nor did he specify what form an independent verification system should take or which international institution should oversee it.
Those questions come as governments at the Human Rights Council confront a more immediate institutional issue: whether to consolidate their work on AI or continue addressing it through separate resolutions and mandates.
The council has considered digital technologies through resolutions and thematic studies for several years. During the current session, experts are also presenting a report examining AI’s effects on cultural rights.
But it has yet to produce unified guidance for governments and technology companies on applying international human rights standards to the development and deployment of AI.
Marc Limon, executive director of the Geneva-based Universal Rights Group, which has advised South Korea on its initiative, has argued that the council needs a dedicated special-procedures mandate to bring its work together.
With additional countries now proposing initiatives, separate efforts risk “muddying the water,” Limon told Geneva Solutions ahead of the session.
“The ultimate goal is to have one set of accessible, unified universal guidance for states on how to regulate AI in a human rights way, and for digital technology companies on how to develop AI that complies with human rights,” he said.
While some governments have resisted proposals for additional human rights oversight of AI on the grounds that regulation could slow technological innovation, Türk said governance also needs to address the technology’s effects on employment practices, fundamental principles of democracy and the environment.