Over 400 scientists warn of serious risks posed by advanced AI models and urge stronger containment measures.
On Tuesday, AI godfathers Geoffrey Hinton, Yoshua Bengio and Stuart Jonathan Russell, accompanied by more than 400 leading scientists, called for the creation of an international body to establish common standards for artificial intelligence (AI) systems, oversee their operation and record all control and containment failures.
RELATED:
Trump Rejects International Regulation of Artificial Intelligence
In an open letter, scientists from more than 30 countries warned of the serious cybersecurity risks associated with the current development of artificial intelligence, following incidents in which several advanced models managed to escape sandboxed environments and compromise external infrastructure.
The researchers called for the implementation of more robust security systems, including the complete isolation of some of these systems from the internet.
The scientists warned that the security failures already recorded could be considered “irresponsible” negligence under standard cybersecurity practices. They noted that such incidents could be prevented through isolated environments and more robust procedures.
Geoffrey Hinton, who is known as the godfather of AI, used his 2024 Nobel Prize acceptance speech not to celebrate his creation but to warn humanity of its risks:
“AI is already being used by authoritarian governments for mass surveillance. In the near future, AI will be used to… pic.twitter.com/WOXrz0v6Ds
— Power to the People ☭🕊 (@ProudSocialist) May 27, 2026
“Without immediate changes to AI development and deployment practices, these incidents will continue to occur, and future ones will be far more damaging to our exposed critical infrastructure,” the researchers stated in their letter.
They reiterated the need for more research into AI safety and greater emphasis on containment technologies and procedures.
Organizations that develop, deploy and test AI models with cyber capabilities should immediately adopt high-security containment systems and completely isolate their models, the researchers said. They also called for the creation of an international body to standardize best practices, which should be required to record and track all containment failures that occur.
The researchers who signed the open letter stressed that, to ensure AI safety and containment research can learn from mistakes, security incidents involving containment failures should be reported to the body and subsequently made public. “This transparency is necessary both for learning and for accountability,” the scientists emphasized.
teleSUR/ JF
Sources: EFE
From teleSUR English via This RSS Feed.