Google Chrome has been found to quietly install a 4 GB local large language model, known as weights.bin, on users' computers, occupying significant disk space without explicit consent. This file is stored in the OptGuideOnDeviceModel folder and is reinstated by Chrome if manually deleted. The discovery, announced by Alexander Hanff, highlights the substantial storage requirements of this model, with considerable climate implications at a billion-device scale1. The silent installation of such a large file raises concerns about data storage and potential security risks associated with local large language models. As Google continues to develop and integrate LLMs into its products, the security implications and risk surfaces of these technologies are becoming increasingly important to consider. The installation of large files without user consent may have significant consequences for device security and management, so what matters most to practitioners is understanding the potential security trade-offs of these emerging technologies.