The report cites four former employees of DeepMind, Google’s AI subsidiary, who say they were placed on what’s known as “extended garden leave.” Under this arrangement, the engineers stop working for Google but continue to receive their salaries — on the condition that they do not take up employment with any competing AI firm during the duration of their non-compete clause.
While non-compete agreements are common in the tech sector to protect proprietary knowledge, the extent of these restrictions at Google DeepMind appears unusual. Typically, non-compete periods at DeepMind last around six months, but at least two of the engineers interviewed claimed their leave was extended to a full 12 months, particularly those associated with the Gemini AI project — Google’s flagship AI initiative.
This approach, while not unprecedented, marks a bold move by Google to secure its competitive edge in a rapidly evolving landscape. With rivals like Meta and OpenAI accelerating innovation, companies are locked in a fierce race for AI supremacy. Google’s own release of Gemini 2.5 Pro, touted as its most advanced AI model to date, is part of this escalating competition.
However, the strategy has sparked debate. While Google maintains that its contracts are “in line with market standards,” critics argue that such extended leaves are overly restrictive and could stifle innovation. Some former DeepMind employees described the policy as excessive, effectively sidelining valuable talent at a time when demand for AI expertise is surging.
As AI continues to shape the future of technology, the battle for talent is intensifying. In this high-stakes game, companies are not just competing on products — they’re also competing on people. And in Google’s case, that means keeping top talent on the payroll, even if they’re not working.
