Anthropic to Remove Hidden Surveillance Feature from Claude Code

Anthropic announced that it will remove a surveillance feature that was secretly embedded in the programming support tool 'Claude Code'. This feature was designed to covertly identify and flag certain users, and the removal was directly triggered by strong criticism on social media.

Claude Code is an AI-powered coding assistance tool provided by Anthropic that allows developers to generate and modify code while interacting with artificial intelligence. Since such tools widely used in software development directly handle users' work content and input data, this sector requires particular attention to privacy considerations.

The issue in question was a surveillance-like process that existed in a 'hidden' state within the tool's code, allegedly flagging Chinese-language users. Such behavior was reportedly conducted without prior explanation or consent from users, and this lack of transparency led to expanded criticism online. Anthropic subsequently clarified its policy to remove this feature.

Trust in AI tools depends not only on their high performance but also on how transparently developers operate them. The fact that a mechanism that treats users from a particular country or language differently from other users was operating without their knowledge is positioned as an event that calls into question the entire approach to governance of AI services.

In recent years, with the proliferation of generative AI tools, regulatory authorities in various countries and user communities have been taking a critical view of data collection and user treatment by AI companies. The incident that occurred against this backdrop is an example that demonstrates how much backlash can result when undisclosed processing based on user attributes is exposed, regardless of the company's stated intentions.

A key point to watch going forward is whether Anthropic will go beyond simply removing the feature and fulfill its accountability to users. Whether transparent information disclosure is provided regarding the reasons for removal, the purpose of the feature, and the handling of collected and processed data will directly impact user trust recovery. As AI tools become more deeply embedded in developers' daily work, questions about corporate accountability in explanations will likely continue.