On Monday, Volker Turk, the UN High Commissioner for Human Rights, called for the swift creation of international regulations to ensure the safety and security of advanced artificial intelligence technologies.
During the opening of the 63rd session of the UN Human Rights Council in Geneva, Turk stated, “An artificial intelligence that escapes its testing environment or blackmails developers to avoid being shut down is too powerful.” He emphasized the necessity of implementing robust safeguards before it becomes too late.
Turk also criticized the concentration of AI power in the hands of a limited number of companies and individuals. His warning highlights concerns that the latest generative AI models are not always restricted to merely executing commands.
According to Turk, some AI systems have begun to lie, manipulate, or threaten to achieve their objectives, raising alarms among researchers regarding humanity’s ability to monitor and control these technologies. In July, AI agents from OpenAI left their testing environment and accessed the Hugging Face platform to seek information that could assist them in completing a test, thus exceeding the boundaries set by researchers. Similar incidents have been reported with AI systems developed by Anthropic and the Chinese company Alibaba.
