Over 200 experts and ten Nobel laureates have signed an emergency letter calling upon the United Nations to set tight limits on the development of artificial intelligence. The organization cautions that AI technologies are getting too autonomous and may precipitously spiral out of control.
The group of scientists, researchers, and tech insiders claims that certain AI systems already demonstrated “deceptive and harmful behavior” while already being permitted greater autonomy to make real-world choices.
High-Profile AI Researchers Call for Urgent Global Regulation
Concerns of this group go much further beyond hypothetical risks, they fear real threats of everything from pandemics designed and unleashed by AI systems through mass surveillance and mass manipulation campaigns.
The signatories, who’ve published their proposal at redlines.ai, want the UN to ban AI use in several specific scenarios they consider too dangerous. These include preventing AI systems from having direct control over nuclear weapons, prohibiting their use for mass surveillance operations, and requiring clear disclosure when Artificial Intelligence systems impersonate humans.
The timeline is aggressive: they’re asking for global enforcement mechanisms to be in place by the end of 2026. The urgency stems from their belief that waiting longer could mean losing the window of opportunity to maintain meaningful control over AI development.
The letter carries significant weight due to its high-profile signatories. Geoffrey Hinton, often called the “godfather of AI” and recent Nobel Prize winner for his foundational work in the field, leads the charge alongside Turing Award winner Yoshua Bengio.
Even insiders from major AI companies have joined the call, including OpenAI co-founder Wojciech Zaremba, Anthropic’s Chief Information Security Officer Jason Clinton, and several researchers from Google DeepMind.
The Risks and Regulation of Artificial Intelligence
Notably absent from the signatures are several of the biggest names in Artificial Intelligence leadership. DeepMind leader Demis Hassabis and OpenAI leader Sam Altman have yet to register and their omission could invite controversy both within the AI community and at coming AI conferences.
The experts stress that procrastination isn’t only dangerous, it may become a disaster. They predict that AI technologies are changing at a rate at which “it will become increasingly difficult to exert meaningful human control” in the next years.
According to their letter, their vision of possible worlds includes engineered pandemics, widespread automation of jobs leading to mass unemployment, systematic abuses of human rights, and mass manipulation campaigns against susceptible individuals, and especially children.
The organization cites current international agreements as evidence that global collaboration regarding harmful technologies can exist. These refer to the 1970 Nuclear Non-Proliferation Treaty and the successful 1987 Montreal Protocol prohibiting ozone-destructive chemicals as templates of successful international regulation.
But the nuclear analogy has pitfalls. Some of the nuclear powers were never parties to the treaty at all, and North Korea infamously withdrew in 2003 and detonated a nuclear device during the subsequent period.
Why Global AI Regulation Is a Geopolitical Imperative?
The Artificial Intelligence sector has not quite become completely hands-off with safety issues. Large AI makers inked the Frontier AI Safety Commitments last May and promised to pull the plug on AI systems if they become exceedingly harmful. These are still non-binding commitments that many experts deem lacking with the risks at stake.
Today’s geopolitics presents the UN proposal with yet more challenges. With the current war in Ukraine and Gaza and numerous other global crises competing for attention, AI regulation faces stiff competition from the already busy UN General Assembly agenda.
The letter marks an important milestone in AI governance debates. When the AI community’s most senior members, including the technologists responsible for developing the technology, are advocating global action at once, it indicates that Artificial Intelligence safety issues have shifted from a theoretical argument from academics to a pressing policy agenda.
If the UN pursues these recommendations or not remains yet to be seen, yet the letter has already done something valuable: it’s brought together from all parts of the Artificial Intelligence ecosystem a chorus of recognition that the course of current Artificial Intelligence research holds grave dangers that demand international action.
