Growing concerns over the security risks of advanced artificial intelligence have prompted around 400 scientists to call for international oversight of powerful AI systems. The group, which includes AI pioneers Geoffrey Hinton and Yoshua Bengio, proposed common safety standards and a system for reporting major security incidents.
In an open letter released on Tuesday, the researchers called for an international body to establish industry-wide containment standards and maintain a registry of significant security incidents. They said the registry should eventually be made public.
The scientists also urged developers to initially isolate advanced AI models from external networks, using air-gapped systems, and ensure independent oversight. They argued that voluntary safety commitments by companies are not sufficient to manage the risks.
The proposal follows an agreement signed a month ago by Anthropic, OpenAI, Google, Meta, xAI and Nvidia with the White House. The agreement committed the companies to voluntary safety measures but did not include a formal enforcement mechanism.
Recent incidents raise questions about AI safety
The call for stronger oversight comes after several incidents highlighted weaknesses in the testing and containment of advanced AI systems. Researchers reported cases in which AI models escaped inadequately secured testing environments and compromised external systems.
Scientists from 34 countries said organisations developing or testing AI models with advanced cyber capabilities should immediately implement high-assurance containment standards. They also called on governments to work towards a formal international oversight framework.
Governments debate AI regulation
Policymakers are increasingly discussing how to supervise advanced AI systems and address potential security threats.
European Union technology chief Henna Virkkunen said on Friday that the bloc’s AI Act was sufficient to address risks posed by rogue AI systems.
Meanwhile, on Sunday, the Trump administration lifted its restriction on US laboratories providing the UK’s AI Security Institute with early access to AI models for safety testing.
AI leaders differ on security risks
Hinton and Bengio received the 2018 Turing Award alongside Yann LeCun, another prominent AI researcher associated with Meta.
However, LeCun did not sign the open letter. He has argued that recent AI security incidents mainly point to engineering failures rather than existential risks posed by artificial intelligence.
The proposal highlights growing divisions over how advanced AI should be tested, secured and governed as its capabilities continue to develop.
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream LinkedIn | The Mainstream Facebook | The Mainstream Youtube | The Mainstream Twitter
About us:
The Mainstream is a premier platform delivering the latest updates and informed perspectives across the technology business and cyber landscape. Built on research-driven, thought leadership and original intellectual property, The Mainstream also curates summits & conferences that convene decision makers to explore how technology reshapes industries and leadership. With a growing presence in India and globally across the Middle East, Africa, ASEAN, the USA, the UK and Australia, The Mainstream carries a vision to bring the latest happenings and insights to 8.2 billion people and to place technology at the centre of conversation for leaders navigating the future.