Google DeepMind cofounder Shane Legg warns that advancing AI must never run ahead of safety and launches the DeepMind Institute to explore the deployment of AGI (Financial Times)

Google DeepMind co-founder Shane Legg has joined the safety-first camp in the AI debate, warning that artificial intelligence advances must not outpace AI safety measures, the Financial Times reported on September 16. With AI capabilities advancing rapidly, Legg cautioned that the field must ensure safety is not left behind. His comments came as Silicon Valley considers slowing the release of new AI models amid company concerns over safety risks.

Anthropic CEO and co-founder Dario Amodei recently published an essay advocating for a more cautious approach to AI development, titled "We Must Pace the Frontier." While Legg found the essay "interesting directionally" and "worth considering," he stressed that more work is needed to evaluate practical implementation. Legg recently launched the DeepMind Institute to explore the implications of artificial general intelligence (AGI), a system that matches human cognitive capabilities.

He noted that AGI is on the horizon and emphasized the need for a deeper understanding of its implications. Despite some executives at OpenAI and Nvidia announcing AGI progress, Legg remained cautious, expressing a "comfortable" 50% chance of minimal AGI emergence by 2028. Meanwhile, Meta CEO Mark Zuckerberg weighed in on the AI safety debate, stating that trust and alignment are becoming the most important capabilities in AI.

He emphasized that labs have a strong incentive to ensure model alignment to prevent potential harm, citing Meta's decision to delay the release of its Muse AI for safety and security reasons.