AI News Summary — Week 25.2, 2026

This week saw a major talent shift in the AI industry, with one of Google's top researchers jumping ship to OpenAI. Meanwhile, world leaders gathered at the G7 to discuss AI governance. Here's the rundown.

Google's Gemini Co-Lead Jumps to OpenAI

In a significant blow to Google, Noam Shazeer — co-lead of the Gemini team and a co-author of the landmark 2017 "Attention Is All You Need" paper that introduced the transformer architecture — is leaving Google to join OpenAI.

The move is particularly striking given that Google paid approximately $2.7 billion to bring Shazeer back less than two years ago, along with his team from Character.AI, the startup he had co-founded after originally leaving Google.

"After less than two years and a multibillion-dollar effort to bring him home, Google's Gemini co-lead is walking across the street," noted The Next Web. "The departure also reframes a debate inside the industry about whether enormous retention packages actually work."

Shazeer is widely credited as a principal architect of Google's Gemini models. His decision to join OpenAI — just as the company heads toward a potential IPO — is being seen as both a research gain and a market signal about where top talent sees the future.

For Google, the loss comes amid intense competition with OpenAI and Anthropic across the enterprise AI stack. The company has been pushing its models into high-profile deployments, including a Pentagon rollout that scaled from tens of thousands to over a million users in just six months.

G7 Leaders Push for US-Led AI Coalition

At the G7 summit in Évian, France, world leaders met with top AI CEOs including Sam Altman (OpenAI), Dario Amodei (Anthropic), and Demis Hassabis (Google DeepMind) to discuss AI governance and keeping advanced AI out of China's hands.

The tech leaders cautioned that frontier models are "too dangerous to ignore" and that AI governance can't be left to tech companies alone. Reports suggest discussions centered on creating a "trusted partners" framework that would give allied nations access to cutting-edge US AI models while explicitly excluding China.

The summit took place just days after the US government forced Anthropic to restrict access to its most powerful models, a move that has sparked widespread criticism from security researchers.

European leaders have been cautious about publicly confronting the Trump administration over the Anthropic decision, with EU officials preferring to address the matter through diplomatic channels rather than direct confrontation at the summit.

Sources: The Next Web, TechCrunch, POLITICO, Capacity