‘We're entering a new age’: Google DeepMind CEO Demis Hassabis on emergence of AGI

Google DeepMind CEO Demis Hassabis warns that we’re running out of time to prepare for artificial general intelligence (AGI), and that humanity must act now to prepare for its arrival. In a short essay shared on X, titled “A Framework for Frontier AI and the Dawning of a New Age,” Hassabis expressed concerns surrounding risks of powerful AI systems and their need for governance. 

“This is a pivotal moment in human history,” Hassabis said, highlighting that AGI is “a few short years away.” “We have to navigate this critical period of development thoughtfully and carefully. Urgent action is needed to address risks that might arise as we get closer to AGI,” he added.

Must read: Anthropic introduces India pricing for Claude AI plans; Pricing starts at…

AGI risks and preparedness

In the essay, Hassabis highlighted the growing concerns around frontier AI models, saying that it has already started to raise concerns for cyberattacks, such as helping hackers write malicious code, identify software vulnerabilities, or automate phishing campaigns. It has also shown nuclear and bio risks, which may occur in the future as AI capabilities improve.

As a result, he stressed that powerful AI will need preparedness and greater control, and the government must bring strong safety measures, regulations, and technical controls to keep these systems safe. 

“We will need robust safeguards to maintain control of increasingly agentic, recursively self-improving systems - and tackle unknown issues that will only become clearer over time,” he added.

Hassabis proposes frontier AI standards body

Hassabis also proposed a US-led coalition, consisting of experts who will test frontier AI models, evaluate safety standards, and decide if they are safe enough to be released to the public. This showcases his approval of the Trump Administration’s recent intervention on the release of AI models by OpenAI and Anthropic.

Must read: OpenAI's first AI device isn't a phone; Its a moving smart speaker to rival Amazon Alexa

“The rapid progress we're seeing in AI requires a new approach to testing frontier AI model capabilities that is dynamic, adaptable, and rigorous. The US is well positioned, given its economic and technical standing, to take the first step in developing such a framework.”

He further highlighted the government-backed coalition should be an independent, nonprofit body that will oversee the safety of advanced AI models. He says it could be modelled after FINRA, the U.S. financial industry's self-regulatory watchdog. He proposed that it could be funded by AI companies themselves, while operating independently to set standards and assess the safety of frontier AI systems.

Several tech leaders, including Google CEO Sundar Pichai, OpenAI’s Sam Altman, Block’s Jack Dorsey, and others, have showcased support for the vision.