Google AI Chief says company has ‘all the ingredients’ to beat China’s DeepSeek

Google’s Response to DeepSeek

At a company-wide meeting in Paris on Wednesday, Google executives addressed questions about DeepSeek, a Chinese AI start-up whose app recently overtook ChatGPT at the top of Apple’s App Store. Alphabet CEO Sundar Pichai read an employee-submitted query asking about the implications of DeepSeek’s fast rise and its claim of training a powerful AI model at a much lower cost than other major players.

Demis Hassabis, CEO of Google DeepMind, downplayed DeepSeek’s claims, suggesting that its reported training expenses likely did not reflect the true scale of the resources used. He also indicated that DeepSeek probably relied heavily on Western AI models and hardware, while stressing that Google’s own models are both more efficient and more powerful. Nonetheless, Hassabis acknowledged that DeepSeek is a strong team worth taking seriously, especially given the security and geopolitical factors involved. U.S. agencies have already restricted employees from using the Chinese AI tool due to security concerns.

Changes to Google’s AI Principles

In the same meeting, Pichai read questions regarding Google’s recent modifications to its AI Principles—particularly the removal of a prior commitment not to use AI for weapons or surveillance. Kent Walker, the company’s president of global affairs, responded by noting that the original guidelines were written in 2018, in a very different technological and global context. He said the new approach allows for more nuanced discussions around how AI is employed in various scenarios.

Google’s original AI Principles came after the company chose not to renew its work on Project Maven, a government program using AI to interpret drone video data. Walker explained that since those early rules were formulated, AI technology has evolved in complexity and scope, leading Google to update its policies to better address the sophisticated challenges that AI now presents.