[DigitalToday reporter Chi-gyu Hwang] Demis Hassabis (데미스 하사비스), CEO of Google DeepMind, proposed setting up an international standards body ahead of the arrival of artificial general intelligence, or AGI.

In a recent post on social media platform X (Twitter), he proposed the idea, saying, "The closer we get to AGI, the greater the risks become."

He proposed establishing a "frontier AI standards body" modeled on the U.S. Financial Industry Regulatory Authority (FINRA). It would be a public-private partnership under federal government oversight, with an independent technical expert and representatives from the open-source community included on its board. Most funding would come from the industry.

Models that pass the criteria set by the standards body would be classified as "frontier-grade," and the organization would be designated a "frontier lab." They would be advised to follow best practices such as disclosing model cards, strengthening internal security and allocating resources for safety research.

Initially, labs would voluntarily submit models for review up to 30 days before release. Once the system stabilizes, the procedure could be made mandatory. Assessments would cover cybersecurity, biological threats, attempts to bypass safeguards and signs of deception. Evaluation items would be updated regularly, and the standards body would also need its own non-public testing capability.

Hassabis said, "This framework applies to frontier-grade models regardless of the country developing them or whether they are open or closed," adding that if the United States moves first, it could be a starting point for setting a shared international standard.

He said AGI has the potential to become a tool that drives advances in science and medicine. But even if technical challenges are solved, society will face questions about new economic models, values and the meaning of life.

He said, "These questions cannot be decided by technologists alone. Everyone in society must step up together. The future has not yet been written. We must use this time before AGI arrives to make the technology benefit all of humanity." He added, "Frontier models have already revealed problems in cybersecurity. If capabilities keep advancing, nuclear and biological risks could soon emerge. Right now we are trapped in intense commercial and geopolitical competition. The pace of frontier technology development is outstripping our pace of understanding," and stressed that cautious optimism is the right strategy.