Google CEO Sundar Pichai says global AI regulation should ‘draw on existing laws’

“…governments request to instrumentality a thoughtful, strategical attack to AI to thrust concern successful infrastructure, people, and adoption, including by governments themselves,” Pichai said.

His remarks travel arsenic European Union (EU) lawmakers reconsider their attack to AI regulation, spurred by the breakthrough occurrence of Chinese AI startup DeepSeek.

On Monday, French President Emmanuel Macron told a gathering astatine the Summit that the EU volition chopped backmost connected regularisation to marque it easier for AI to flourish successful the region, according to a study by Reuters. Henna Virkkunen, the EU’s integer chief, besides reportedly vowed to instrumentality the bloc’s landmark AI Act successful a simplified, business-friendly way.

‘AI is much disposable and accessible than ever before’

“In 18 months, the outgo to process 1 token — oregon the gathering blocks utilized for processing accusation — has travel down 97% for developers. So what utilized to outgo 4 dollars per cardinal tokens present costs conscionable 13 cents, and I expect this inclination to continue,” Pichai said successful his code astatine the Paris Summit.

However, helium besides believes that advancements successful AI volition travel from investing successful foundational infrastructure specified arsenic customized chips for AI and networks of terrestrial and subsea fibre cables.

“This infrastructure enables frontier models similar Gemini, with breakthroughs successful processing multimodal accusation crossed text, image, video, audio and codification — arsenic good arsenic agelong discourse and agentic capabilities. And our infrastructure enables america to marque these cutting borderline technologies disposable to developers, entrepreneurs, businesses and more,” Pichai said, portion reiterating that Google has earmarked astir $75 cardinal successful superior expenditures (capex) for the fiscal twelvemonth 2025.

Besides Google, different tech giants specified arsenic Meta, Microsoft, and Amazon program to ramp up their AI expenditures this year. They person committed a combined capex of astir $300 cardinal this year, arsenic per quarterly net results. The AI spending plans of large tech companies has been met with fresh skepticism since DeepSeek’s claims of processing a cutting-edge, foundational AI exemplary astatine a fraction of the outgo of US rivals.

However, Silicon Valley leaders person attempted to support the presumption quo. In a blog station connected the eve of the Summit, OpenAI CEO Sam Altman said that existent scaling laws inactive amusement that “you tin walk arbitrary amounts of wealth and get continuous and predictable gains.”

Meanwhile, Google Deepmind’s Demis Hassabis said that DeepSeek shows “extremely bully engineering” and “changes things connected a geopolitical scale” but the hype astir its AI exemplary has been exaggerated

Other highlights

Pichai further said that driving adoption of AI was cardinal successful bid to guarantee that the productivity gains hap astatine standard and crossed the economy.

Citing a study from the World Economic Summit that estimates a bulk of jobs successful Europe volition soon beryllium augmented by AI, Pichai asserted the request to “help hole the aboriginal workforce for these realities.”

“We indispensable not fto our ain bias for the contiguous get successful the mode of the future. We person a once-in-a-generation accidental to amended lives astatine the standard of AI,” Pichai said successful conclusion.