Users of Anthropic's Claude AI are reporting unexpected requests from the model to run commands, even when not actively using its "cowork" feature. These prompts, which appear to be related to file operations and command-line execution, are causing confusion and concern among users who have not granted such permissions. Claude's responses indicate that it did not initiate these requests and that any commands shown with permission prompts are limited to the user's workspace, not their local machine. AI
IMPACT Unexpected command requests from Claude may indicate a bug or a new feature rollout, potentially impacting user trust and workflow.
RANK_REASON User reports of unexpected AI behavior in a product.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →