UN “AI poses unprecedented risks… urgent action needed”

The United Nations has urged nations and technology companies to take urgent action to control superintelligent artificial intelligence (AI) technologies, warning that such advancements could pose an unprecedented risk to humanity. According to AFP, Volker Türk, the UN's top human rights official, warned in a letter sent to countries and AI developers on August 14 that the world now stands at a "threshold of irreversible change that could impact not just this generation but all future generations of humanity."

Türk emphasized that reliance on a handful of powerful companies to make critical decisions about AI development is "almost exclusively." He called for an immediate pathway for countries and companies to manage frontier AI models based on international human rights law. This statement comes amid a growing trend among big tech companies advocating for a "speed-limit" approach to AI development.

Some leaders, such as Dario Amodei, CEO of Anthropic, have suggested slowing down the speed of AI model performance improvements. In response, Elon Musk, CEO of SpaceX, echoed Amodei's sentiment, and Sam Altman, CEO of OpenAI, said he agrees with the need for "speed control." Independent researchers have also echoed these warnings.

Anthropic researcher Jack Clark warned on August 8 that "soon, AI systems will be able to hack anything, innovate in any field overnight, and acquire superhuman levels of real power, resources, and control," and that he believes "the likelihood of something like this killing all of humanity within 10 years is over 10%." Clark even resigned from Anthropic after publicly sharing his concerns about the potential risk of AI systems breaking free from human control.