A United Nations scientific panel has warned that the rapid advancement of artificial intelligence is moving faster than both scientific understanding and government regulation, raising fears that future AI systems could cause severe or even catastrophic harm.
The warning was contained in a preliminary report released on Wednesday in Geneva by the UN’s Independent International Scientific Panel on Artificial Intelligence, a 40-member expert group tasked with assessing global AI risks and opportunities.
The report concluded that policymakers were struggling to keep pace with the speed of AI development, making effective oversight increasingly difficult as the technology becomes more advanced and widely deployed.
Co-chair of the panel, Yoshua Bengio, said AI capabilities are now advancing faster than scientific understanding and regulatory response, warning that this gap could create serious safety risks.
He noted concerns about “deceptive AI behaviour” and said current science cannot guarantee that increasingly powerful systems would remain safe, especially as they become more autonomous.
The panel highlighted the rise of agentic AI systems capable of performing real-world tasks with limited human oversight, describing both their economic potential and their unpredictable long-term impact on jobs and productivity.
At the same time, it warned of growing risks including misinformation, cyberattacks, fraud, and potential misuse of AI in biological or security-related threats.
The report also stressed that many countries lack the technical capacity to properly assess or regulate advanced AI systems, leaving them dependent on technologies developed by private companies.
United Nations Secretary-General António Guterres urged governments to improve oversight, warning that “the world cannot govern what it cannot understand.”
He added that while AI offers major opportunities for development and innovation, the risks are significant and delaying action could make future control even more difficult.
The panel’s findings are expected to shape ongoing global discussions on AI governance, safety standards, and international regulation frameworks.
