Hassabis pitched AI oversight body before DeepMind's 4% overhaul

Demis Hassabis pitched a new AI-oversight entity to Trump administration officials including Treasury Secretary Scott Bessent, days before Alphabet's DeepMind overhaul triggered a 4 percent stock drop.

"Brin addressed hundreds of employees in an April town hall, urging the company's Google DeepMind AI lab to move faster," one person familiar with the remarks said, as Anthropic raced ahead with a preview of its Claude Mythos model.

On August 5, Alphabet announced the sweeping leadership overhaul. Hassabis stepped aside as DeepMind chief in favor of his deputy Koray Kavukcuoglu, becoming chair with fewer management obligations. The two original technical co-leads of Gemini, Jeff Dean and Oriol Vinyals, quit to co-found a startup. Google's stock fell 4 percent the day of the announcement.

The changes mark a further erosion of DeepMind's autonomy since Google acquired the London-based lab in 2014, as Alphabet pushes its AI to power more software and generate more revenue. Kavukcuoglu now has final say on major DeepMind decisions, and several nontechnical teams are moving into corporate Google's reporting structure.

The oversight body proposal, which has not been previously reported, comes as the AI industry faces intensifying regulatory scrutiny. Hassabis discussed the proposed entity with heads of other AI labs and administration officials, according to people familiar with the matter. The proposal indicates a potential shift toward formalized AI governance structures that could affect major tech firms including OpenAI, Anthropic, and Meta.

Google's Gemini model briefly overtook competitors last November with the release of Gemini 3, but new updates from Anthropic and OpenAI put it back in catch-up mode. By August, Google had delayed the new flagship version of Gemini by two months after internal testing showed its performance continued to lag behind rivals in areas such as coding.

Brin, who no longer holds an executive title, has been informally influencing model training efforts. He has pushed resource allocation toward specific areas such as recursive self-improvement — the point at which the technology can improve without human intervention. The 52-year-old co-founder has used his implicit power to fire up staff, something he also did after OpenAI's ChatGPT launch in 2022 jolted Google.

Power Shifts From London to Mountain View

Kavukcuoglu has been consolidating power since moving from London to Mountain View last year for the chief AI architect promotion, reporting directly to CEO Sundar Pichai. Some London-based leaders found their influence on Gemini's technical direction diminished as Hassabis decreased his involvement and Kavukcuoglu became the central figure shaping the model's direction.

Even before his relocation, Kavukcuoglu was the primary point of contact between DeepMind and Google Cloud, the company's main AI revenue engine. Senior leaders in the cloud division expect his appointment to alleviate tensions with DeepMind over allocation of Google's constrained supply of Tensor Processing Units (TPUs).

Structural Challenges Persist

Four sources characterized Google's laggard status as a result of lingering structural challenges. Constraints in computing capacity and disagreements between Gemini's many project leaders led to fewer resources for critical areas such as coding. Google's bureaucratic nature also caused it to release models at a slower pace than nimbler rivals.

Pichai told Wall Street analysts last month that the company would continue to scale up its infrastructure to alleviate constraints around TPUs needed for model training, as well as consumer and enterprise products.

Alphabet shares, which trade at roughly 22 times forward earnings, fell 4 percent on the day of the announcement. The restructuring shows that Google is prioritizing commercial AI deployment over pure research, a shift that could accelerate Gemini's integration across Google's product suite but may also raise concerns about the erosion of DeepMind's research culture. With Anthropic's Claude Mythos and OpenAI's latest models setting the pace, investors will be watching whether Kavukcuoglu's more commercially focused leadership can close the gap in coding and other critical benchmarks.

This article is for informational purposes only and does not constitute investment advice.