Google’s internal AI split
- Reports say Google has created a divide where some engineers get broader access to internal AI tools than others. - Access appears to vary by team and the sensitivity of work, creating tiers of permission internally. - Uneven AI access implies roles trusted with production systems may gain more long‑term value and possibly compensation premium (timesofindia.indiatimes.com).
Google has reportedly split AI access inside its engineering ranks, with some teams allowed to use Anthropic’s Claude while most others are steered to Google’s own Gemini tools. (businessinsider.com) Business Insider reported on April 21 that some Google DeepMind engineers have Claude access for coding, while many engineers elsewhere at Google do not. The report said the difference tracks team assignments and the sensitivity of the systems employees work on. (businessinsider.com) That split lands after Google spent the past year pushing employees to use AI more aggressively in daily work. Google said in August 2025 that AI was generating 30% of new code at the company and contributing to an estimated 10% increase in engineering velocity. (blog.google) Google has also tightened the boundary around outside tools. Multiple reports in 2025 said software engineers were told to rely on internal AI systems for coding and to get approval before using third-party tools for other tasks. (9to5google.com, news18.com) The practical issue is not just who gets an assistant, but which assistant they get. Bloomberg reported April 21 that Google leaders are worried about lagging rivals in AI coding tools, even as competitors such as Anthropic are gaining traction with developers and businesses. (bloomberg.com) Inside Google, that can turn tool access into a status marker. Engineers working on production systems or in research groups trusted with more sensitive work can end up with broader permissions, while others are expected to prove productivity gains on the company standard stack. (businessinsider.com, timesofindia.indiatimes.com) Google executives have pushed back on the idea that the company is broadly behind on internal AI adoption. VentureBeat reported last week that Google DeepMind Chief Executive Demis Hassabis and other leaders disputed claims that only a small slice of engineers are using advanced AI tools. (venturebeat.com) Alphabet is making that argument from a position of scale. The company reported its fourth-quarter and full-year 2025 results on February 4, 2026, after a year in which AI spending and product rollouts remained central to its business narrative. (abc.xyz, blog.google) The immediate question is whether Google keeps those internal AI tiers in place or broadens access beyond DeepMind. Either way, the company’s own push to make AI a core job skill has made tool access part of how engineering work is divided and judged. (businessinsider.com, blog.google)
