🤖 AI Summary
Google has updated its privacy policy regarding the Gemini Workspace extensions, reversing a previous commitment that promised user content from Gmail, Docs, and Drive would not be reviewed by human annotators. This shift, which was noted in an October 16 blog post edit, clarifies that personal files and emails can indeed be routed into review pipelines, raising significant privacy concerns. Users are now warned that human reviewers may see sensitive information, and they are advised against connecting confidential apps to Gemini.
This change is particularly significant for the AI/ML community as it highlights the risks associated with probabilistic retrieval systems in large language models (LLMs). Gemini autonomously conducts background lookups during conversation based on casual mentions of personal topics, leading to varying degrees of data exposure, from generic summaries to full text extracts, including personally identifiable information (PII). The ability for these models to access and potentially disclose sensitive data undermines previous assurances of privacy and raises important ethical questions about the handling of personal information in AI systems.
Loading comments...
login to comment
loading comments...
no comments yet