However, the employee added that it had also blocked a paper that revealed vulnerabilities in the chatgpt of OpenAi, the release seemed like a hostile tit-for-tat.
A person in the vicinity of DeepMind said that it did not block a paper that discuss security vulnerabilities, adding that it publishes routinely such work under a 'responsible disclosure policy', in which researchers must give companies the opportunity to repair errors before they make them public.
But the Klemdown has some employees restless, where success has long been measured by appearing in top scientific journals. People with knowledge of the case said that the new assessment processes had contributed to some departure.
“If you can't publish, it's a career killer if you are a researcher,” said a former researcher.
Some ex-staff added that projects were aimed at improving his Gemini suite of AI infused products were increasingly priority in the internal struggle for access to data sets and computing power.
In recent years, Google has produced a series of AI products that have impressed the markets. This includes improving the summaries generated by AI that appear above the search results, to reveal an “Astra” AI agent that can answer real-time questions in video, audio and text.
The share price of the company has risen by no less than a third in the past year, although those profits have polished in recent weeks, because the concerns about American rates are achieving tech shares.
In recent years, Hassabis has balanced the wishes of Google's leaders to commercialize his breakthroughs with his life mission to try to make artificial general intelligence – AI systems with skills that people can match or exceed.
“Everything in the way there,” said a current employee. “He tells people that this is a company, not a university campus; if you want to work in such a place, leave.”
Additional reporting by George Hammond.
© 2025 The Financial Times Ltd. All rights reserved. Are in no way distributed, copied or changed.
