A new fault line has emerged inside Google: The Claude haves and have-nots

A new conflict has surfaced within Google, highlighting a stark division among its employees regarding access to Anthropic's Claude AI tool. Reports indicate that select members of the Google DeepMind team have recently been granted permission to utilize Claude for coding tasks. This development has raised eyebrows, as the general policy restricts employees from using any tools that are not developed by Google or customized for internal purposes. Claude has rapidly gained popularity as one of the leading AI coding tools in the tech industry. However, the decision to allow certain DeepMind engineers to access it has sparked frustration among other Google engineers, who are confined to using the company’s own Gemini AI models. Many of these employees believe that the internal models lag behind Claude in terms of performance. The growing reliance on AI within Google has led to increased expectations for employees to integrate AI into their work processes. Some engineers have been assigned specific AI-related objectives that will influence their performance reviews this year. They are not only expected to generate code using AI but also to develop tools that enhance their workflow efficiency. A spokesperson for Google did not provide a comment regarding these developments when approached by Business Insider. The company typically keeps its engineers tied to internal tools for several reasons, including the custom nature of its infrastructure and the belief that using their own products can lead to quicker improvements. In contrast, other tech firms, like Meta, permit their employees to access external AI models such as Claude. Recently, Steve Yegge, a programmer and blogger, shared insights on X regarding Google's AI progress, suggesting that it lags behind other industries. His remarks prompted a response from Demis Hassabis, CEO of Google DeepMind, who strongly refuted Yegge's claims, labeling them as misleading. Despite Hassabis's denial, Yegge later indicated that some Google employees had confirmed his observations, including the fact that while DeepMind engineers could use Claude, others in the company could not. An internal discussion about equal access reportedly proposed revoking Claude access for everyone, but DeepMind staff objected so vehemently that some threatened to leave the company. This situation underscores the ongoing tensions and challenges within Google as it navigates the integration of AI technologies across its divisions.