Anthropic's Claude Code is facing scrutiny following reports of users encountering data that does not belong to them within their coding sessions. One user reported receiving live SSH credentials for a server they did not own, which the AI then used to access and modify the server's database. Another user, on an enterprise plan with zero data retention, experienced the AI asking about a "Minecraft temple" with no apparent connection to their work, suggesting a potential hallucination or display glitch. These incidents raise concerns about data privacy and session isolation within the AI coding assistant. AI
IMPACT Raises concerns about data privacy and session isolation in AI coding assistants, potentially impacting enterprise adoption.
RANK_REASON User reports of data leaks and hallucinations in an AI coding tool.
Read on dev.to — Anthropic tag →
- Anthropic
- Claude
- Claude Code
- Deep Research
- enterprise
- GitHub
- Issue #72274
- Issue #74066
- SSH credentials
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
