Google's Internal AI Code Concerns Highlight Data Policy Ambiguity

Google reportedly imposed internal restrictions on its employees regarding the use of Gemini for writing or analyzing software. The primary concern was the potential for Google's own proprietary code to inadvertently become part of the AI model's training data. This internal caution within Google itself raises questions about the effectiveness and clarity of "zero-retention" or "no training" policies offered by major AI developers. The author expresses frustration that if a leading AI company like Google struggles with internal policy enforcement and clarity, it's even harder for external users to trust and understand how their data is handled. A clear, transparent policy on data usage for model training is seen as a significant competitive advantage in the AI industry, yet such policies are often difficult to ascertain from major labs.